1 /*
   2  * Copyright (c) 1997, 2016, Oracle and/or its affiliates. All rights reserved.
   3  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   4  *
   5  * This code is free software; you can redistribute it and/or modify it
   6  * under the terms of the GNU General Public License version 2 only, as
   7  * published by the Free Software Foundation.
   8  *
   9  * This code is distributed in the hope that it will be useful, but WITHOUT
  10  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
  11  * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  12  * version 2 for more details (a copy is included in the LICENSE file that
  13  * accompanied this code).
  14  *
  15  * You should have received a copy of the GNU General Public License version
  16  * 2 along with this work; if not, write to the Free Software Foundation,
  17  * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
  18  *
  19  * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
  20  * or visit www.oracle.com if you need additional information or have any
  21  * questions.
  22  *
  23  */
  24 
  25 #include "precompiled.hpp"
  26 #include "asm/assembler.hpp"
  27 #include "asm/assembler.inline.hpp"
  28 #include "gc/shared/cardTableModRefBS.hpp"
  29 #include "gc/shared/collectedHeap.inline.hpp"
  30 #include "interpreter/interpreter.hpp"
  31 #include "memory/resourceArea.hpp"
  32 #include "prims/methodHandles.hpp"
  33 #include "runtime/biasedLocking.hpp"
  34 #include "runtime/interfaceSupport.hpp"
  35 #include "runtime/objectMonitor.hpp"
  36 #include "runtime/os.hpp"
  37 #include "runtime/sharedRuntime.hpp"
  38 #include "runtime/stubRoutines.hpp"
  39 #include "utilities/macros.hpp"
  40 #if INCLUDE_ALL_GCS
  41 #include "gc/g1/g1CollectedHeap.inline.hpp"
  42 #include "gc/g1/g1SATBCardTableModRefBS.hpp"
  43 #include "gc/g1/heapRegion.hpp"
  44 #endif // INCLUDE_ALL_GCS
  45 
  46 #ifdef PRODUCT
  47 #define BLOCK_COMMENT(str) /* nothing */
  48 #define STOP(error) stop(error)
  49 #else
  50 #define BLOCK_COMMENT(str) block_comment(str)
  51 #define STOP(error) block_comment(error); stop(error)
  52 #endif
  53 
  54 #define BIND(label) bind(label); BLOCK_COMMENT(#label ":")
  55 // Implementation of AddressLiteral
  56 
  57 // A 2-D table for managing compressed displacement(disp8) on EVEX enabled platforms.
  58 unsigned char tuple_table[Assembler::EVEX_ETUP + 1][Assembler::AVX_512bit + 1] = {
  59   // -----------------Table 4.5 -------------------- //
  60   16, 32, 64,  // EVEX_FV(0)
  61   4,  4,  4,   // EVEX_FV(1) - with Evex.b
  62   16, 32, 64,  // EVEX_FV(2) - with Evex.w
  63   8,  8,  8,   // EVEX_FV(3) - with Evex.w and Evex.b
  64   8,  16, 32,  // EVEX_HV(0)
  65   4,  4,  4,   // EVEX_HV(1) - with Evex.b
  66   // -----------------Table 4.6 -------------------- //
  67   16, 32, 64,  // EVEX_FVM(0)
  68   1,  1,  1,   // EVEX_T1S(0)
  69   2,  2,  2,   // EVEX_T1S(1)
  70   4,  4,  4,   // EVEX_T1S(2)
  71   8,  8,  8,   // EVEX_T1S(3)
  72   4,  4,  4,   // EVEX_T1F(0)
  73   8,  8,  8,   // EVEX_T1F(1)
  74   8,  8,  8,   // EVEX_T2(0)
  75   0,  16, 16,  // EVEX_T2(1)
  76   0,  16, 16,  // EVEX_T4(0)
  77   0,  0,  32,  // EVEX_T4(1)
  78   0,  0,  32,  // EVEX_T8(0)
  79   8,  16, 32,  // EVEX_HVM(0)
  80   4,  8,  16,  // EVEX_QVM(0)
  81   2,  4,  8,   // EVEX_OVM(0)
  82   16, 16, 16,  // EVEX_M128(0)
  83   8,  32, 64,  // EVEX_DUP(0)
  84   0,  0,  0    // EVEX_NTUP
  85 };
  86 
  87 AddressLiteral::AddressLiteral(address target, relocInfo::relocType rtype) {
  88   _is_lval = false;
  89   _target = target;
  90   switch (rtype) {
  91   case relocInfo::oop_type:
  92   case relocInfo::metadata_type:
  93     // Oops are a special case. Normally they would be their own section
  94     // but in cases like icBuffer they are literals in the code stream that
  95     // we don't have a section for. We use none so that we get a literal address
  96     // which is always patchable.
  97     break;
  98   case relocInfo::external_word_type:
  99     _rspec = external_word_Relocation::spec(target);
 100     break;
 101   case relocInfo::internal_word_type:
 102     _rspec = internal_word_Relocation::spec(target);
 103     break;
 104   case relocInfo::opt_virtual_call_type:
 105     _rspec = opt_virtual_call_Relocation::spec();
 106     break;
 107   case relocInfo::static_call_type:
 108     _rspec = static_call_Relocation::spec();
 109     break;
 110   case relocInfo::runtime_call_type:
 111     _rspec = runtime_call_Relocation::spec();
 112     break;
 113   case relocInfo::poll_type:
 114   case relocInfo::poll_return_type:
 115     _rspec = Relocation::spec_simple(rtype);
 116     break;
 117   case relocInfo::none:
 118     break;
 119   default:
 120     ShouldNotReachHere();
 121     break;
 122   }
 123 }
 124 
 125 // Implementation of Address
 126 
 127 #ifdef _LP64
 128 
 129 Address Address::make_array(ArrayAddress adr) {
 130   // Not implementable on 64bit machines
 131   // Should have been handled higher up the call chain.
 132   ShouldNotReachHere();
 133   return Address();
 134 }
 135 
 136 // exceedingly dangerous constructor
 137 Address::Address(int disp, address loc, relocInfo::relocType rtype) {
 138   _base  = noreg;
 139   _index = noreg;
 140   _scale = no_scale;
 141   _disp  = disp;
 142   switch (rtype) {
 143     case relocInfo::external_word_type:
 144       _rspec = external_word_Relocation::spec(loc);
 145       break;
 146     case relocInfo::internal_word_type:
 147       _rspec = internal_word_Relocation::spec(loc);
 148       break;
 149     case relocInfo::runtime_call_type:
 150       // HMM
 151       _rspec = runtime_call_Relocation::spec();
 152       break;
 153     case relocInfo::poll_type:
 154     case relocInfo::poll_return_type:
 155       _rspec = Relocation::spec_simple(rtype);
 156       break;
 157     case relocInfo::none:
 158       break;
 159     default:
 160       ShouldNotReachHere();
 161   }
 162 }
 163 #else // LP64
 164 
 165 Address Address::make_array(ArrayAddress adr) {
 166   AddressLiteral base = adr.base();
 167   Address index = adr.index();
 168   assert(index._disp == 0, "must not have disp"); // maybe it can?
 169   Address array(index._base, index._index, index._scale, (intptr_t) base.target());
 170   array._rspec = base._rspec;
 171   return array;
 172 }
 173 
 174 // exceedingly dangerous constructor
 175 Address::Address(address loc, RelocationHolder spec) {
 176   _base  = noreg;
 177   _index = noreg;
 178   _scale = no_scale;
 179   _disp  = (intptr_t) loc;
 180   _rspec = spec;
 181 }
 182 
 183 #endif // _LP64
 184 
 185 
 186 
 187 // Convert the raw encoding form into the form expected by the constructor for
 188 // Address.  An index of 4 (rsp) corresponds to having no index, so convert
 189 // that to noreg for the Address constructor.
 190 Address Address::make_raw(int base, int index, int scale, int disp, relocInfo::relocType disp_reloc) {
 191   RelocationHolder rspec;
 192   if (disp_reloc != relocInfo::none) {
 193     rspec = Relocation::spec_simple(disp_reloc);
 194   }
 195   bool valid_index = index != rsp->encoding();
 196   if (valid_index) {
 197     Address madr(as_Register(base), as_Register(index), (Address::ScaleFactor)scale, in_ByteSize(disp));
 198     madr._rspec = rspec;
 199     return madr;
 200   } else {
 201     Address madr(as_Register(base), noreg, Address::no_scale, in_ByteSize(disp));
 202     madr._rspec = rspec;
 203     return madr;
 204   }
 205 }
 206 
 207 // Implementation of Assembler
 208 
 209 int AbstractAssembler::code_fill_byte() {
 210   return (u_char)'\xF4'; // hlt
 211 }
 212 
 213 // make this go away someday
 214 void Assembler::emit_data(jint data, relocInfo::relocType rtype, int format) {
 215   if (rtype == relocInfo::none)
 216     emit_int32(data);
 217   else
 218     emit_data(data, Relocation::spec_simple(rtype), format);
 219 }
 220 
 221 void Assembler::emit_data(jint data, RelocationHolder const& rspec, int format) {
 222   assert(imm_operand == 0, "default format must be immediate in this file");
 223   assert(inst_mark() != NULL, "must be inside InstructionMark");
 224   if (rspec.type() !=  relocInfo::none) {
 225     #ifdef ASSERT
 226       check_relocation(rspec, format);
 227     #endif
 228     // Do not use AbstractAssembler::relocate, which is not intended for
 229     // embedded words.  Instead, relocate to the enclosing instruction.
 230 
 231     // hack. call32 is too wide for mask so use disp32
 232     if (format == call32_operand)
 233       code_section()->relocate(inst_mark(), rspec, disp32_operand);
 234     else
 235       code_section()->relocate(inst_mark(), rspec, format);
 236   }
 237   emit_int32(data);
 238 }
 239 
 240 static int encode(Register r) {
 241   int enc = r->encoding();
 242   if (enc >= 8) {
 243     enc -= 8;
 244   }
 245   return enc;
 246 }
 247 
 248 void Assembler::emit_arith_b(int op1, int op2, Register dst, int imm8) {
 249   assert(dst->has_byte_register(), "must have byte register");
 250   assert(isByte(op1) && isByte(op2), "wrong opcode");
 251   assert(isByte(imm8), "not a byte");
 252   assert((op1 & 0x01) == 0, "should be 8bit operation");
 253   emit_int8(op1);
 254   emit_int8(op2 | encode(dst));
 255   emit_int8(imm8);
 256 }
 257 
 258 
 259 void Assembler::emit_arith(int op1, int op2, Register dst, int32_t imm32) {
 260   assert(isByte(op1) && isByte(op2), "wrong opcode");
 261   assert((op1 & 0x01) == 1, "should be 32bit operation");
 262   assert((op1 & 0x02) == 0, "sign-extension bit should not be set");
 263   if (is8bit(imm32)) {
 264     emit_int8(op1 | 0x02); // set sign bit
 265     emit_int8(op2 | encode(dst));
 266     emit_int8(imm32 & 0xFF);
 267   } else {
 268     emit_int8(op1);
 269     emit_int8(op2 | encode(dst));
 270     emit_int32(imm32);
 271   }
 272 }
 273 
 274 // Force generation of a 4 byte immediate value even if it fits into 8bit
 275 void Assembler::emit_arith_imm32(int op1, int op2, Register dst, int32_t imm32) {
 276   assert(isByte(op1) && isByte(op2), "wrong opcode");
 277   assert((op1 & 0x01) == 1, "should be 32bit operation");
 278   assert((op1 & 0x02) == 0, "sign-extension bit should not be set");
 279   emit_int8(op1);
 280   emit_int8(op2 | encode(dst));
 281   emit_int32(imm32);
 282 }
 283 
 284 // immediate-to-memory forms
 285 void Assembler::emit_arith_operand(int op1, Register rm, Address adr, int32_t imm32) {
 286   assert((op1 & 0x01) == 1, "should be 32bit operation");
 287   assert((op1 & 0x02) == 0, "sign-extension bit should not be set");
 288   if (is8bit(imm32)) {
 289     emit_int8(op1 | 0x02); // set sign bit
 290     emit_operand(rm, adr, 1);
 291     emit_int8(imm32 & 0xFF);
 292   } else {
 293     emit_int8(op1);
 294     emit_operand(rm, adr, 4);
 295     emit_int32(imm32);
 296   }
 297 }
 298 
 299 
 300 void Assembler::emit_arith(int op1, int op2, Register dst, Register src) {
 301   assert(isByte(op1) && isByte(op2), "wrong opcode");
 302   emit_int8(op1);
 303   emit_int8(op2 | encode(dst) << 3 | encode(src));
 304 }
 305 
 306 
 307 bool Assembler::query_compressed_disp_byte(int disp, bool is_evex_inst, int vector_len,
 308                                            int cur_tuple_type, int in_size_in_bits, int cur_encoding) {
 309   int mod_idx = 0;
 310   // We will test if the displacement fits the compressed format and if so
 311   // apply the compression to the displacment iff the result is8bit.
 312   if (VM_Version::supports_evex() && is_evex_inst) {
 313     switch (cur_tuple_type) {
 314     case EVEX_FV:
 315       if ((cur_encoding & VEX_W) == VEX_W) {
 316         mod_idx = ((cur_encoding & EVEX_Rb) == EVEX_Rb) ? 3 : 2;
 317       } else {
 318         mod_idx = ((cur_encoding & EVEX_Rb) == EVEX_Rb) ? 1 : 0;
 319       }
 320       break;
 321 
 322     case EVEX_HV:
 323       mod_idx = ((cur_encoding & EVEX_Rb) == EVEX_Rb) ? 1 : 0;
 324       break;
 325 
 326     case EVEX_FVM:
 327       break;
 328 
 329     case EVEX_T1S:
 330       switch (in_size_in_bits) {
 331       case EVEX_8bit:
 332         break;
 333 
 334       case EVEX_16bit:
 335         mod_idx = 1;
 336         break;
 337 
 338       case EVEX_32bit:
 339         mod_idx = 2;
 340         break;
 341 
 342       case EVEX_64bit:
 343         mod_idx = 3;
 344         break;
 345       }
 346       break;
 347 
 348     case EVEX_T1F:
 349     case EVEX_T2:
 350     case EVEX_T4:
 351       mod_idx = (in_size_in_bits == EVEX_64bit) ? 1 : 0;
 352       break;
 353 
 354     case EVEX_T8:
 355       break;
 356 
 357     case EVEX_HVM:
 358       break;
 359 
 360     case EVEX_QVM:
 361       break;
 362 
 363     case EVEX_OVM:
 364       break;
 365 
 366     case EVEX_M128:
 367       break;
 368 
 369     case EVEX_DUP:
 370       break;
 371 
 372     default:
 373       assert(0, "no valid evex tuple_table entry");
 374       break;
 375     }
 376 
 377     if (vector_len >= AVX_128bit && vector_len <= AVX_512bit) {
 378       int disp_factor = tuple_table[cur_tuple_type + mod_idx][vector_len];
 379       if ((disp % disp_factor) == 0) {
 380         int new_disp = disp / disp_factor;
 381         if ((-0x80 <= new_disp && new_disp < 0x80)) {
 382           disp = new_disp;
 383         }
 384       } else {
 385         return false;
 386       }
 387     }
 388   }
 389   return (-0x80 <= disp && disp < 0x80);
 390 }
 391 
 392 
 393 bool Assembler::emit_compressed_disp_byte(int &disp) {
 394   int mod_idx = 0;
 395   // We will test if the displacement fits the compressed format and if so
 396   // apply the compression to the displacment iff the result is8bit.
 397   if (VM_Version::supports_evex() && _attributes && _attributes->is_evex_instruction()) {
 398     int evex_encoding = _attributes->get_evex_encoding();
 399     int tuple_type = _attributes->get_tuple_type();
 400     switch (tuple_type) {
 401     case EVEX_FV:
 402       if ((evex_encoding & VEX_W) == VEX_W) {
 403         mod_idx = ((evex_encoding & EVEX_Rb) == EVEX_Rb) ? 3 : 2;
 404       } else {
 405         mod_idx = ((evex_encoding & EVEX_Rb) == EVEX_Rb) ? 1 : 0;
 406       }
 407       break;
 408 
 409     case EVEX_HV:
 410       mod_idx = ((evex_encoding & EVEX_Rb) == EVEX_Rb) ? 1 : 0;
 411       break;
 412 
 413     case EVEX_FVM:
 414       break;
 415 
 416     case EVEX_T1S:
 417       switch (_attributes->get_input_size()) {
 418       case EVEX_8bit:
 419         break;
 420 
 421       case EVEX_16bit:
 422         mod_idx = 1;
 423         break;
 424 
 425       case EVEX_32bit:
 426         mod_idx = 2;
 427         break;
 428 
 429       case EVEX_64bit:
 430         mod_idx = 3;
 431         break;
 432       }
 433       break;
 434 
 435     case EVEX_T1F:
 436     case EVEX_T2:
 437     case EVEX_T4:
 438       mod_idx = (_attributes->get_input_size() == EVEX_64bit) ? 1 : 0;
 439       break;
 440 
 441     case EVEX_T8:
 442       break;
 443 
 444     case EVEX_HVM:
 445       break;
 446 
 447     case EVEX_QVM:
 448       break;
 449 
 450     case EVEX_OVM:
 451       break;
 452 
 453     case EVEX_M128:
 454       break;
 455 
 456     case EVEX_DUP:
 457       break;
 458 
 459     default:
 460       assert(0, "no valid evex tuple_table entry");
 461       break;
 462     }
 463 
 464     int vector_len = _attributes->get_vector_len();
 465     if (vector_len >= AVX_128bit && vector_len <= AVX_512bit) {
 466       int disp_factor = tuple_table[tuple_type + mod_idx][vector_len];
 467       if ((disp % disp_factor) == 0) {
 468         int new_disp = disp / disp_factor;
 469         if (is8bit(new_disp)) {
 470           disp = new_disp;
 471         }
 472       } else {
 473         return false;
 474       }
 475     }
 476   }
 477   return is8bit(disp);
 478 }
 479 
 480 
 481 void Assembler::emit_operand(Register reg, Register base, Register index,
 482                              Address::ScaleFactor scale, int disp,
 483                              RelocationHolder const& rspec,
 484                              int rip_relative_correction) {
 485   relocInfo::relocType rtype = (relocInfo::relocType) rspec.type();
 486 
 487   // Encode the registers as needed in the fields they are used in
 488 
 489   int regenc = encode(reg) << 3;
 490   int indexenc = index->is_valid() ? encode(index) << 3 : 0;
 491   int baseenc = base->is_valid() ? encode(base) : 0;
 492 
 493   if (base->is_valid()) {
 494     if (index->is_valid()) {
 495       assert(scale != Address::no_scale, "inconsistent address");
 496       // [base + index*scale + disp]
 497       if (disp == 0 && rtype == relocInfo::none  &&
 498           base != rbp LP64_ONLY(&& base != r13)) {
 499         // [base + index*scale]
 500         // [00 reg 100][ss index base]
 501         assert(index != rsp, "illegal addressing mode");
 502         emit_int8(0x04 | regenc);
 503         emit_int8(scale << 6 | indexenc | baseenc);
 504       } else if (emit_compressed_disp_byte(disp) && rtype == relocInfo::none) {
 505         // [base + index*scale + imm8]
 506         // [01 reg 100][ss index base] imm8
 507         assert(index != rsp, "illegal addressing mode");
 508         emit_int8(0x44 | regenc);
 509         emit_int8(scale << 6 | indexenc | baseenc);
 510         emit_int8(disp & 0xFF);
 511       } else {
 512         // [base + index*scale + disp32]
 513         // [10 reg 100][ss index base] disp32
 514         assert(index != rsp, "illegal addressing mode");
 515         emit_int8(0x84 | regenc);
 516         emit_int8(scale << 6 | indexenc | baseenc);
 517         emit_data(disp, rspec, disp32_operand);
 518       }
 519     } else if (base == rsp LP64_ONLY(|| base == r12)) {
 520       // [rsp + disp]
 521       if (disp == 0 && rtype == relocInfo::none) {
 522         // [rsp]
 523         // [00 reg 100][00 100 100]
 524         emit_int8(0x04 | regenc);
 525         emit_int8(0x24);
 526       } else if (emit_compressed_disp_byte(disp) && rtype == relocInfo::none) {
 527         // [rsp + imm8]
 528         // [01 reg 100][00 100 100] disp8
 529         emit_int8(0x44 | regenc);
 530         emit_int8(0x24);
 531         emit_int8(disp & 0xFF);
 532       } else {
 533         // [rsp + imm32]
 534         // [10 reg 100][00 100 100] disp32
 535         emit_int8(0x84 | regenc);
 536         emit_int8(0x24);
 537         emit_data(disp, rspec, disp32_operand);
 538       }
 539     } else {
 540       // [base + disp]
 541       assert(base != rsp LP64_ONLY(&& base != r12), "illegal addressing mode");
 542       if (disp == 0 && rtype == relocInfo::none &&
 543           base != rbp LP64_ONLY(&& base != r13)) {
 544         // [base]
 545         // [00 reg base]
 546         emit_int8(0x00 | regenc | baseenc);
 547       } else if (emit_compressed_disp_byte(disp) && rtype == relocInfo::none) {
 548         // [base + disp8]
 549         // [01 reg base] disp8
 550         emit_int8(0x40 | regenc | baseenc);
 551         emit_int8(disp & 0xFF);
 552       } else {
 553         // [base + disp32]
 554         // [10 reg base] disp32
 555         emit_int8(0x80 | regenc | baseenc);
 556         emit_data(disp, rspec, disp32_operand);
 557       }
 558     }
 559   } else {
 560     if (index->is_valid()) {
 561       assert(scale != Address::no_scale, "inconsistent address");
 562       // [index*scale + disp]
 563       // [00 reg 100][ss index 101] disp32
 564       assert(index != rsp, "illegal addressing mode");
 565       emit_int8(0x04 | regenc);
 566       emit_int8(scale << 6 | indexenc | 0x05);
 567       emit_data(disp, rspec, disp32_operand);
 568     } else if (rtype != relocInfo::none ) {
 569       // [disp] (64bit) RIP-RELATIVE (32bit) abs
 570       // [00 000 101] disp32
 571 
 572       emit_int8(0x05 | regenc);
 573       // Note that the RIP-rel. correction applies to the generated
 574       // disp field, but _not_ to the target address in the rspec.
 575 
 576       // disp was created by converting the target address minus the pc
 577       // at the start of the instruction. That needs more correction here.
 578       // intptr_t disp = target - next_ip;
 579       assert(inst_mark() != NULL, "must be inside InstructionMark");
 580       address next_ip = pc() + sizeof(int32_t) + rip_relative_correction;
 581       int64_t adjusted = disp;
 582       // Do rip-rel adjustment for 64bit
 583       LP64_ONLY(adjusted -=  (next_ip - inst_mark()));
 584       assert(is_simm32(adjusted),
 585              "must be 32bit offset (RIP relative address)");
 586       emit_data((int32_t) adjusted, rspec, disp32_operand);
 587 
 588     } else {
 589       // 32bit never did this, did everything as the rip-rel/disp code above
 590       // [disp] ABSOLUTE
 591       // [00 reg 100][00 100 101] disp32
 592       emit_int8(0x04 | regenc);
 593       emit_int8(0x25);
 594       emit_data(disp, rspec, disp32_operand);
 595     }
 596   }
 597 }
 598 
 599 void Assembler::emit_operand(XMMRegister reg, Register base, Register index,
 600                              Address::ScaleFactor scale, int disp,
 601                              RelocationHolder const& rspec) {
 602   if (UseAVX > 2) {
 603     int xreg_enc = reg->encoding();
 604     if (xreg_enc > 15) {
 605       XMMRegister new_reg = as_XMMRegister(xreg_enc & 0xf);
 606       emit_operand((Register)new_reg, base, index, scale, disp, rspec);
 607       return;
 608     }
 609   }
 610   emit_operand((Register)reg, base, index, scale, disp, rspec);
 611 }
 612 
 613 // Secret local extension to Assembler::WhichOperand:
 614 #define end_pc_operand (_WhichOperand_limit)
 615 
 616 address Assembler::locate_operand(address inst, WhichOperand which) {
 617   // Decode the given instruction, and return the address of
 618   // an embedded 32-bit operand word.
 619 
 620   // If "which" is disp32_operand, selects the displacement portion
 621   // of an effective address specifier.
 622   // If "which" is imm64_operand, selects the trailing immediate constant.
 623   // If "which" is call32_operand, selects the displacement of a call or jump.
 624   // Caller is responsible for ensuring that there is such an operand,
 625   // and that it is 32/64 bits wide.
 626 
 627   // If "which" is end_pc_operand, find the end of the instruction.
 628 
 629   address ip = inst;
 630   bool is_64bit = false;
 631 
 632   debug_only(bool has_disp32 = false);
 633   int tail_size = 0; // other random bytes (#32, #16, etc.) at end of insn
 634 
 635   again_after_prefix:
 636   switch (0xFF & *ip++) {
 637 
 638   // These convenience macros generate groups of "case" labels for the switch.
 639 #define REP4(x) (x)+0: case (x)+1: case (x)+2: case (x)+3
 640 #define REP8(x) (x)+0: case (x)+1: case (x)+2: case (x)+3: \
 641              case (x)+4: case (x)+5: case (x)+6: case (x)+7
 642 #define REP16(x) REP8((x)+0): \
 643               case REP8((x)+8)
 644 
 645   case CS_segment:
 646   case SS_segment:
 647   case DS_segment:
 648   case ES_segment:
 649   case FS_segment:
 650   case GS_segment:
 651     // Seems dubious
 652     LP64_ONLY(assert(false, "shouldn't have that prefix"));
 653     assert(ip == inst+1, "only one prefix allowed");
 654     goto again_after_prefix;
 655 
 656   case 0x67:
 657   case REX:
 658   case REX_B:
 659   case REX_X:
 660   case REX_XB:
 661   case REX_R:
 662   case REX_RB:
 663   case REX_RX:
 664   case REX_RXB:
 665     NOT_LP64(assert(false, "64bit prefixes"));
 666     goto again_after_prefix;
 667 
 668   case REX_W:
 669   case REX_WB:
 670   case REX_WX:
 671   case REX_WXB:
 672   case REX_WR:
 673   case REX_WRB:
 674   case REX_WRX:
 675   case REX_WRXB:
 676     NOT_LP64(assert(false, "64bit prefixes"));
 677     is_64bit = true;
 678     goto again_after_prefix;
 679 
 680   case 0xFF: // pushq a; decl a; incl a; call a; jmp a
 681   case 0x88: // movb a, r
 682   case 0x89: // movl a, r
 683   case 0x8A: // movb r, a
 684   case 0x8B: // movl r, a
 685   case 0x8F: // popl a
 686     debug_only(has_disp32 = true);
 687     break;
 688 
 689   case 0x68: // pushq #32
 690     if (which == end_pc_operand) {
 691       return ip + 4;
 692     }
 693     assert(which == imm_operand && !is_64bit, "pushl has no disp32 or 64bit immediate");
 694     return ip;                  // not produced by emit_operand
 695 
 696   case 0x66: // movw ... (size prefix)
 697     again_after_size_prefix2:
 698     switch (0xFF & *ip++) {
 699     case REX:
 700     case REX_B:
 701     case REX_X:
 702     case REX_XB:
 703     case REX_R:
 704     case REX_RB:
 705     case REX_RX:
 706     case REX_RXB:
 707     case REX_W:
 708     case REX_WB:
 709     case REX_WX:
 710     case REX_WXB:
 711     case REX_WR:
 712     case REX_WRB:
 713     case REX_WRX:
 714     case REX_WRXB:
 715       NOT_LP64(assert(false, "64bit prefix found"));
 716       goto again_after_size_prefix2;
 717     case 0x8B: // movw r, a
 718     case 0x89: // movw a, r
 719       debug_only(has_disp32 = true);
 720       break;
 721     case 0xC7: // movw a, #16
 722       debug_only(has_disp32 = true);
 723       tail_size = 2;  // the imm16
 724       break;
 725     case 0x0F: // several SSE/SSE2 variants
 726       ip--;    // reparse the 0x0F
 727       goto again_after_prefix;
 728     default:
 729       ShouldNotReachHere();
 730     }
 731     break;
 732 
 733   case REP8(0xB8): // movl/q r, #32/#64(oop?)
 734     if (which == end_pc_operand)  return ip + (is_64bit ? 8 : 4);
 735     // these asserts are somewhat nonsensical
 736 #ifndef _LP64
 737     assert(which == imm_operand || which == disp32_operand,
 738            "which %d is_64_bit %d ip " INTPTR_FORMAT, which, is_64bit, p2i(ip));
 739 #else
 740     assert((which == call32_operand || which == imm_operand) && is_64bit ||
 741            which == narrow_oop_operand && !is_64bit,
 742            "which %d is_64_bit %d ip " INTPTR_FORMAT, which, is_64bit, p2i(ip));
 743 #endif // _LP64
 744     return ip;
 745 
 746   case 0x69: // imul r, a, #32
 747   case 0xC7: // movl a, #32(oop?)
 748     tail_size = 4;
 749     debug_only(has_disp32 = true); // has both kinds of operands!
 750     break;
 751 
 752   case 0x0F: // movx..., etc.
 753     switch (0xFF & *ip++) {
 754     case 0x3A: // pcmpestri
 755       tail_size = 1;
 756     case 0x38: // ptest, pmovzxbw
 757       ip++; // skip opcode
 758       debug_only(has_disp32 = true); // has both kinds of operands!
 759       break;
 760 
 761     case 0x70: // pshufd r, r/a, #8
 762       debug_only(has_disp32 = true); // has both kinds of operands!
 763     case 0x73: // psrldq r, #8
 764       tail_size = 1;
 765       break;
 766 
 767     case 0x12: // movlps
 768     case 0x28: // movaps
 769     case 0x2E: // ucomiss
 770     case 0x2F: // comiss
 771     case 0x54: // andps
 772     case 0x55: // andnps
 773     case 0x56: // orps
 774     case 0x57: // xorps
 775     case 0x58: // addpd
 776     case 0x59: // mulpd
 777     case 0x6E: // movd
 778     case 0x7E: // movd
 779     case 0xAE: // ldmxcsr, stmxcsr, fxrstor, fxsave, clflush
 780     case 0xFE: // paddd
 781       debug_only(has_disp32 = true);
 782       break;
 783 
 784     case 0xAD: // shrd r, a, %cl
 785     case 0xAF: // imul r, a
 786     case 0xBE: // movsbl r, a (movsxb)
 787     case 0xBF: // movswl r, a (movsxw)
 788     case 0xB6: // movzbl r, a (movzxb)
 789     case 0xB7: // movzwl r, a (movzxw)
 790     case REP16(0x40): // cmovl cc, r, a
 791     case 0xB0: // cmpxchgb
 792     case 0xB1: // cmpxchg
 793     case 0xC1: // xaddl
 794     case 0xC7: // cmpxchg8
 795     case REP16(0x90): // setcc a
 796       debug_only(has_disp32 = true);
 797       // fall out of the switch to decode the address
 798       break;
 799 
 800     case 0xC4: // pinsrw r, a, #8
 801       debug_only(has_disp32 = true);
 802     case 0xC5: // pextrw r, r, #8
 803       tail_size = 1;  // the imm8
 804       break;
 805 
 806     case 0xAC: // shrd r, a, #8
 807       debug_only(has_disp32 = true);
 808       tail_size = 1;  // the imm8
 809       break;
 810 
 811     case REP16(0x80): // jcc rdisp32
 812       if (which == end_pc_operand)  return ip + 4;
 813       assert(which == call32_operand, "jcc has no disp32 or imm");
 814       return ip;
 815     default:
 816       ShouldNotReachHere();
 817     }
 818     break;
 819 
 820   case 0x81: // addl a, #32; addl r, #32
 821     // also: orl, adcl, sbbl, andl, subl, xorl, cmpl
 822     // on 32bit in the case of cmpl, the imm might be an oop
 823     tail_size = 4;
 824     debug_only(has_disp32 = true); // has both kinds of operands!
 825     break;
 826 
 827   case 0x83: // addl a, #8; addl r, #8
 828     // also: orl, adcl, sbbl, andl, subl, xorl, cmpl
 829     debug_only(has_disp32 = true); // has both kinds of operands!
 830     tail_size = 1;
 831     break;
 832 
 833   case 0x9B:
 834     switch (0xFF & *ip++) {
 835     case 0xD9: // fnstcw a
 836       debug_only(has_disp32 = true);
 837       break;
 838     default:
 839       ShouldNotReachHere();
 840     }
 841     break;
 842 
 843   case REP4(0x00): // addb a, r; addl a, r; addb r, a; addl r, a
 844   case REP4(0x10): // adc...
 845   case REP4(0x20): // and...
 846   case REP4(0x30): // xor...
 847   case REP4(0x08): // or...
 848   case REP4(0x18): // sbb...
 849   case REP4(0x28): // sub...
 850   case 0xF7: // mull a
 851   case 0x8D: // lea r, a
 852   case 0x87: // xchg r, a
 853   case REP4(0x38): // cmp...
 854   case 0x85: // test r, a
 855     debug_only(has_disp32 = true); // has both kinds of operands!
 856     break;
 857 
 858   case 0xC1: // sal a, #8; sar a, #8; shl a, #8; shr a, #8
 859   case 0xC6: // movb a, #8
 860   case 0x80: // cmpb a, #8
 861   case 0x6B: // imul r, a, #8
 862     debug_only(has_disp32 = true); // has both kinds of operands!
 863     tail_size = 1; // the imm8
 864     break;
 865 
 866   case 0xC4: // VEX_3bytes
 867   case 0xC5: // VEX_2bytes
 868     assert((UseAVX > 0), "shouldn't have VEX prefix");
 869     assert(ip == inst+1, "no prefixes allowed");
 870     // C4 and C5 are also used as opcodes for PINSRW and PEXTRW instructions
 871     // but they have prefix 0x0F and processed when 0x0F processed above.
 872     //
 873     // In 32-bit mode the VEX first byte C4 and C5 alias onto LDS and LES
 874     // instructions (these instructions are not supported in 64-bit mode).
 875     // To distinguish them bits [7:6] are set in the VEX second byte since
 876     // ModRM byte can not be of the form 11xxxxxx in 32-bit mode. To set
 877     // those VEX bits REX and vvvv bits are inverted.
 878     //
 879     // Fortunately C2 doesn't generate these instructions so we don't need
 880     // to check for them in product version.
 881 
 882     // Check second byte
 883     NOT_LP64(assert((0xC0 & *ip) == 0xC0, "shouldn't have LDS and LES instructions"));
 884 
 885     int vex_opcode;
 886     // First byte
 887     if ((0xFF & *inst) == VEX_3bytes) {
 888       vex_opcode = VEX_OPCODE_MASK & *ip;
 889       ip++; // third byte
 890       is_64bit = ((VEX_W & *ip) == VEX_W);
 891     } else {
 892       vex_opcode = VEX_OPCODE_0F;
 893     }
 894     ip++; // opcode
 895     // To find the end of instruction (which == end_pc_operand).
 896     switch (vex_opcode) {
 897       case VEX_OPCODE_0F:
 898         switch (0xFF & *ip) {
 899         case 0x70: // pshufd r, r/a, #8
 900         case 0x71: // ps[rl|ra|ll]w r, #8
 901         case 0x72: // ps[rl|ra|ll]d r, #8
 902         case 0x73: // ps[rl|ra|ll]q r, #8
 903         case 0xC2: // cmp[ps|pd|ss|sd] r, r, r/a, #8
 904         case 0xC4: // pinsrw r, r, r/a, #8
 905         case 0xC5: // pextrw r/a, r, #8
 906         case 0xC6: // shufp[s|d] r, r, r/a, #8
 907           tail_size = 1;  // the imm8
 908           break;
 909         }
 910         break;
 911       case VEX_OPCODE_0F_3A:
 912         tail_size = 1;
 913         break;
 914     }
 915     ip++; // skip opcode
 916     debug_only(has_disp32 = true); // has both kinds of operands!
 917     break;
 918 
 919   case 0x62: // EVEX_4bytes
 920     assert((UseAVX > 0), "shouldn't have EVEX prefix");
 921     assert(ip == inst+1, "no prefixes allowed");
 922     // no EVEX collisions, all instructions that have 0x62 opcodes
 923     // have EVEX versions and are subopcodes of 0x66
 924     ip++; // skip P0 and exmaine W in P1
 925     is_64bit = ((VEX_W & *ip) == VEX_W);
 926     ip++; // move to P2
 927     ip++; // skip P2, move to opcode
 928     // To find the end of instruction (which == end_pc_operand).
 929     switch (0xFF & *ip) {
 930     case 0x22: // pinsrd r, r/a, #8
 931     case 0x61: // pcmpestri r, r/a, #8
 932     case 0x70: // pshufd r, r/a, #8
 933     case 0x73: // psrldq r, #8
 934       tail_size = 1;  // the imm8
 935       break;
 936     default:
 937       break;
 938     }
 939     ip++; // skip opcode
 940     debug_only(has_disp32 = true); // has both kinds of operands!
 941     break;
 942 
 943   case 0xD1: // sal a, 1; sar a, 1; shl a, 1; shr a, 1
 944   case 0xD3: // sal a, %cl; sar a, %cl; shl a, %cl; shr a, %cl
 945   case 0xD9: // fld_s a; fst_s a; fstp_s a; fldcw a
 946   case 0xDD: // fld_d a; fst_d a; fstp_d a
 947   case 0xDB: // fild_s a; fistp_s a; fld_x a; fstp_x a
 948   case 0xDF: // fild_d a; fistp_d a
 949   case 0xD8: // fadd_s a; fsubr_s a; fmul_s a; fdivr_s a; fcomp_s a
 950   case 0xDC: // fadd_d a; fsubr_d a; fmul_d a; fdivr_d a; fcomp_d a
 951   case 0xDE: // faddp_d a; fsubrp_d a; fmulp_d a; fdivrp_d a; fcompp_d a
 952     debug_only(has_disp32 = true);
 953     break;
 954 
 955   case 0xE8: // call rdisp32
 956   case 0xE9: // jmp  rdisp32
 957     if (which == end_pc_operand)  return ip + 4;
 958     assert(which == call32_operand, "call has no disp32 or imm");
 959     return ip;
 960 
 961   case 0xF0:                    // Lock
 962     assert(os::is_MP(), "only on MP");
 963     goto again_after_prefix;
 964 
 965   case 0xF3:                    // For SSE
 966   case 0xF2:                    // For SSE2
 967     switch (0xFF & *ip++) {
 968     case REX:
 969     case REX_B:
 970     case REX_X:
 971     case REX_XB:
 972     case REX_R:
 973     case REX_RB:
 974     case REX_RX:
 975     case REX_RXB:
 976     case REX_W:
 977     case REX_WB:
 978     case REX_WX:
 979     case REX_WXB:
 980     case REX_WR:
 981     case REX_WRB:
 982     case REX_WRX:
 983     case REX_WRXB:
 984       NOT_LP64(assert(false, "found 64bit prefix"));
 985       ip++;
 986     default:
 987       ip++;
 988     }
 989     debug_only(has_disp32 = true); // has both kinds of operands!
 990     break;
 991 
 992   default:
 993     ShouldNotReachHere();
 994 
 995 #undef REP8
 996 #undef REP16
 997   }
 998 
 999   assert(which != call32_operand, "instruction is not a call, jmp, or jcc");
1000 #ifdef _LP64
1001   assert(which != imm_operand, "instruction is not a movq reg, imm64");
1002 #else
1003   // assert(which != imm_operand || has_imm32, "instruction has no imm32 field");
1004   assert(which != imm_operand || has_disp32, "instruction has no imm32 field");
1005 #endif // LP64
1006   assert(which != disp32_operand || has_disp32, "instruction has no disp32 field");
1007 
1008   // parse the output of emit_operand
1009   int op2 = 0xFF & *ip++;
1010   int base = op2 & 0x07;
1011   int op3 = -1;
1012   const int b100 = 4;
1013   const int b101 = 5;
1014   if (base == b100 && (op2 >> 6) != 3) {
1015     op3 = 0xFF & *ip++;
1016     base = op3 & 0x07;   // refetch the base
1017   }
1018   // now ip points at the disp (if any)
1019 
1020   switch (op2 >> 6) {
1021   case 0:
1022     // [00 reg  100][ss index base]
1023     // [00 reg  100][00   100  esp]
1024     // [00 reg base]
1025     // [00 reg  100][ss index  101][disp32]
1026     // [00 reg  101]               [disp32]
1027 
1028     if (base == b101) {
1029       if (which == disp32_operand)
1030         return ip;              // caller wants the disp32
1031       ip += 4;                  // skip the disp32
1032     }
1033     break;
1034 
1035   case 1:
1036     // [01 reg  100][ss index base][disp8]
1037     // [01 reg  100][00   100  esp][disp8]
1038     // [01 reg base]               [disp8]
1039     ip += 1;                    // skip the disp8
1040     break;
1041 
1042   case 2:
1043     // [10 reg  100][ss index base][disp32]
1044     // [10 reg  100][00   100  esp][disp32]
1045     // [10 reg base]               [disp32]
1046     if (which == disp32_operand)
1047       return ip;                // caller wants the disp32
1048     ip += 4;                    // skip the disp32
1049     break;
1050 
1051   case 3:
1052     // [11 reg base]  (not a memory addressing mode)
1053     break;
1054   }
1055 
1056   if (which == end_pc_operand) {
1057     return ip + tail_size;
1058   }
1059 
1060 #ifdef _LP64
1061   assert(which == narrow_oop_operand && !is_64bit, "instruction is not a movl adr, imm32");
1062 #else
1063   assert(which == imm_operand, "instruction has only an imm field");
1064 #endif // LP64
1065   return ip;
1066 }
1067 
1068 address Assembler::locate_next_instruction(address inst) {
1069   // Secretly share code with locate_operand:
1070   return locate_operand(inst, end_pc_operand);
1071 }
1072 
1073 
1074 #ifdef ASSERT
1075 void Assembler::check_relocation(RelocationHolder const& rspec, int format) {
1076   address inst = inst_mark();
1077   assert(inst != NULL && inst < pc(), "must point to beginning of instruction");
1078   address opnd;
1079 
1080   Relocation* r = rspec.reloc();
1081   if (r->type() == relocInfo::none) {
1082     return;
1083   } else if (r->is_call() || format == call32_operand) {
1084     // assert(format == imm32_operand, "cannot specify a nonzero format");
1085     opnd = locate_operand(inst, call32_operand);
1086   } else if (r->is_data()) {
1087     assert(format == imm_operand || format == disp32_operand
1088            LP64_ONLY(|| format == narrow_oop_operand), "format ok");
1089     opnd = locate_operand(inst, (WhichOperand)format);
1090   } else {
1091     assert(format == imm_operand, "cannot specify a format");
1092     return;
1093   }
1094   assert(opnd == pc(), "must put operand where relocs can find it");
1095 }
1096 #endif // ASSERT
1097 
1098 void Assembler::emit_operand32(Register reg, Address adr) {
1099   assert(reg->encoding() < 8, "no extended registers");
1100   assert(!adr.base_needs_rex() && !adr.index_needs_rex(), "no extended registers");
1101   emit_operand(reg, adr._base, adr._index, adr._scale, adr._disp,
1102                adr._rspec);
1103 }
1104 
1105 void Assembler::emit_operand(Register reg, Address adr,
1106                              int rip_relative_correction) {
1107   emit_operand(reg, adr._base, adr._index, adr._scale, adr._disp,
1108                adr._rspec,
1109                rip_relative_correction);
1110 }
1111 
1112 void Assembler::emit_operand(XMMRegister reg, Address adr) {
1113   emit_operand(reg, adr._base, adr._index, adr._scale, adr._disp,
1114                adr._rspec);
1115 }
1116 
1117 // MMX operations
1118 void Assembler::emit_operand(MMXRegister reg, Address adr) {
1119   assert(!adr.base_needs_rex() && !adr.index_needs_rex(), "no extended registers");
1120   emit_operand((Register)reg, adr._base, adr._index, adr._scale, adr._disp, adr._rspec);
1121 }
1122 
1123 // work around gcc (3.2.1-7a) bug
1124 void Assembler::emit_operand(Address adr, MMXRegister reg) {
1125   assert(!adr.base_needs_rex() && !adr.index_needs_rex(), "no extended registers");
1126   emit_operand((Register)reg, adr._base, adr._index, adr._scale, adr._disp, adr._rspec);
1127 }
1128 
1129 
1130 void Assembler::emit_farith(int b1, int b2, int i) {
1131   assert(isByte(b1) && isByte(b2), "wrong opcode");
1132   assert(0 <= i &&  i < 8, "illegal stack offset");
1133   emit_int8(b1);
1134   emit_int8(b2 + i);
1135 }
1136 
1137 
1138 // Now the Assembler instructions (identical for 32/64 bits)
1139 
1140 void Assembler::adcl(Address dst, int32_t imm32) {
1141   InstructionMark im(this);
1142   prefix(dst);
1143   emit_arith_operand(0x81, rdx, dst, imm32);
1144 }
1145 
1146 void Assembler::adcl(Address dst, Register src) {
1147   InstructionMark im(this);
1148   prefix(dst, src);
1149   emit_int8(0x11);
1150   emit_operand(src, dst);
1151 }
1152 
1153 void Assembler::adcl(Register dst, int32_t imm32) {
1154   prefix(dst);
1155   emit_arith(0x81, 0xD0, dst, imm32);
1156 }
1157 
1158 void Assembler::adcl(Register dst, Address src) {
1159   InstructionMark im(this);
1160   prefix(src, dst);
1161   emit_int8(0x13);
1162   emit_operand(dst, src);
1163 }
1164 
1165 void Assembler::adcl(Register dst, Register src) {
1166   (void) prefix_and_encode(dst->encoding(), src->encoding());
1167   emit_arith(0x13, 0xC0, dst, src);
1168 }
1169 
1170 void Assembler::addl(Address dst, int32_t imm32) {
1171   InstructionMark im(this);
1172   prefix(dst);
1173   emit_arith_operand(0x81, rax, dst, imm32);
1174 }
1175 
1176 void Assembler::addl(Address dst, Register src) {
1177   InstructionMark im(this);
1178   prefix(dst, src);
1179   emit_int8(0x01);
1180   emit_operand(src, dst);
1181 }
1182 
1183 void Assembler::addl(Register dst, int32_t imm32) {
1184   prefix(dst);
1185   emit_arith(0x81, 0xC0, dst, imm32);
1186 }
1187 
1188 void Assembler::addl(Register dst, Address src) {
1189   InstructionMark im(this);
1190   prefix(src, dst);
1191   emit_int8(0x03);
1192   emit_operand(dst, src);
1193 }
1194 
1195 void Assembler::addl(Register dst, Register src) {
1196   (void) prefix_and_encode(dst->encoding(), src->encoding());
1197   emit_arith(0x03, 0xC0, dst, src);
1198 }
1199 
1200 void Assembler::addr_nop_4() {
1201   assert(UseAddressNop, "no CPU support");
1202   // 4 bytes: NOP DWORD PTR [EAX+0]
1203   emit_int8(0x0F);
1204   emit_int8(0x1F);
1205   emit_int8(0x40); // emit_rm(cbuf, 0x1, EAX_enc, EAX_enc);
1206   emit_int8(0);    // 8-bits offset (1 byte)
1207 }
1208 
1209 void Assembler::addr_nop_5() {
1210   assert(UseAddressNop, "no CPU support");
1211   // 5 bytes: NOP DWORD PTR [EAX+EAX*0+0] 8-bits offset
1212   emit_int8(0x0F);
1213   emit_int8(0x1F);
1214   emit_int8(0x44); // emit_rm(cbuf, 0x1, EAX_enc, 0x4);
1215   emit_int8(0x00); // emit_rm(cbuf, 0x0, EAX_enc, EAX_enc);
1216   emit_int8(0);    // 8-bits offset (1 byte)
1217 }
1218 
1219 void Assembler::addr_nop_7() {
1220   assert(UseAddressNop, "no CPU support");
1221   // 7 bytes: NOP DWORD PTR [EAX+0] 32-bits offset
1222   emit_int8(0x0F);
1223   emit_int8(0x1F);
1224   emit_int8((unsigned char)0x80);
1225                    // emit_rm(cbuf, 0x2, EAX_enc, EAX_enc);
1226   emit_int32(0);   // 32-bits offset (4 bytes)
1227 }
1228 
1229 void Assembler::addr_nop_8() {
1230   assert(UseAddressNop, "no CPU support");
1231   // 8 bytes: NOP DWORD PTR [EAX+EAX*0+0] 32-bits offset
1232   emit_int8(0x0F);
1233   emit_int8(0x1F);
1234   emit_int8((unsigned char)0x84);
1235                    // emit_rm(cbuf, 0x2, EAX_enc, 0x4);
1236   emit_int8(0x00); // emit_rm(cbuf, 0x0, EAX_enc, EAX_enc);
1237   emit_int32(0);   // 32-bits offset (4 bytes)
1238 }
1239 
1240 void Assembler::addsd(XMMRegister dst, XMMRegister src) {
1241   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
1242   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
1243   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
1244   emit_int8(0x58);
1245   emit_int8((unsigned char)(0xC0 | encode));
1246 }
1247 
1248 void Assembler::addsd(XMMRegister dst, Address src) {
1249   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
1250   InstructionMark im(this);
1251   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
1252   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
1253   simd_prefix(dst, dst, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
1254   emit_int8(0x58);
1255   emit_operand(dst, src);
1256 }
1257 
1258 void Assembler::addss(XMMRegister dst, XMMRegister src) {
1259   NOT_LP64(assert(VM_Version::supports_sse(), ""));
1260   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
1261   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
1262   emit_int8(0x58);
1263   emit_int8((unsigned char)(0xC0 | encode));
1264 }
1265 
1266 void Assembler::addss(XMMRegister dst, Address src) {
1267   NOT_LP64(assert(VM_Version::supports_sse(), ""));
1268   InstructionMark im(this);
1269   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
1270   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
1271   simd_prefix(dst, dst, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
1272   emit_int8(0x58);
1273   emit_operand(dst, src);
1274 }
1275 
1276 void Assembler::aesdec(XMMRegister dst, Address src) {
1277   assert(VM_Version::supports_aes(), "");
1278   InstructionMark im(this);
1279   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
1280   simd_prefix(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
1281   emit_int8((unsigned char)0xDE);
1282   emit_operand(dst, src);
1283 }
1284 
1285 void Assembler::aesdec(XMMRegister dst, XMMRegister src) {
1286   assert(VM_Version::supports_aes(), "");
1287   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
1288   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
1289   emit_int8((unsigned char)0xDE);
1290   emit_int8(0xC0 | encode);
1291 }
1292 
1293 void Assembler::aesdeclast(XMMRegister dst, Address src) {
1294   assert(VM_Version::supports_aes(), "");
1295   InstructionMark im(this);
1296   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
1297   simd_prefix(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
1298   emit_int8((unsigned char)0xDF);
1299   emit_operand(dst, src);
1300 }
1301 
1302 void Assembler::aesdeclast(XMMRegister dst, XMMRegister src) {
1303   assert(VM_Version::supports_aes(), "");
1304   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
1305   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
1306   emit_int8((unsigned char)0xDF);
1307   emit_int8((unsigned char)(0xC0 | encode));
1308 }
1309 
1310 void Assembler::aesenc(XMMRegister dst, Address src) {
1311   assert(VM_Version::supports_aes(), "");
1312   InstructionMark im(this);
1313   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
1314   simd_prefix(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
1315   emit_int8((unsigned char)0xDC);
1316   emit_operand(dst, src);
1317 }
1318 
1319 void Assembler::aesenc(XMMRegister dst, XMMRegister src) {
1320   assert(VM_Version::supports_aes(), "");
1321   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
1322   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
1323   emit_int8((unsigned char)0xDC);
1324   emit_int8(0xC0 | encode);
1325 }
1326 
1327 void Assembler::aesenclast(XMMRegister dst, Address src) {
1328   assert(VM_Version::supports_aes(), "");
1329   InstructionMark im(this);
1330   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
1331   simd_prefix(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
1332   emit_int8((unsigned char)0xDD);
1333   emit_operand(dst, src);
1334 }
1335 
1336 void Assembler::aesenclast(XMMRegister dst, XMMRegister src) {
1337   assert(VM_Version::supports_aes(), "");
1338   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
1339   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
1340   emit_int8((unsigned char)0xDD);
1341   emit_int8((unsigned char)(0xC0 | encode));
1342 }
1343 
1344 void Assembler::andl(Address dst, int32_t imm32) {
1345   InstructionMark im(this);
1346   prefix(dst);
1347   emit_int8((unsigned char)0x81);
1348   emit_operand(rsp, dst, 4);
1349   emit_int32(imm32);
1350 }
1351 
1352 void Assembler::andl(Register dst, int32_t imm32) {
1353   prefix(dst);
1354   emit_arith(0x81, 0xE0, dst, imm32);
1355 }
1356 
1357 void Assembler::andl(Register dst, Address src) {
1358   InstructionMark im(this);
1359   prefix(src, dst);
1360   emit_int8(0x23);
1361   emit_operand(dst, src);
1362 }
1363 
1364 void Assembler::andl(Register dst, Register src) {
1365   (void) prefix_and_encode(dst->encoding(), src->encoding());
1366   emit_arith(0x23, 0xC0, dst, src);
1367 }
1368 
1369 void Assembler::andnl(Register dst, Register src1, Register src2) {
1370   assert(VM_Version::supports_bmi1(), "bit manipulation instructions not supported");
1371   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
1372   int encode = vex_prefix_and_encode(dst->encoding(), src1->encoding(), src2->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
1373   emit_int8((unsigned char)0xF2);
1374   emit_int8((unsigned char)(0xC0 | encode));
1375 }
1376 
1377 void Assembler::andnl(Register dst, Register src1, Address src2) {
1378   assert(VM_Version::supports_bmi1(), "bit manipulation instructions not supported");
1379   InstructionMark im(this);
1380   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
1381   vex_prefix(src2, src1->encoding(), dst->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
1382   emit_int8((unsigned char)0xF2);
1383   emit_operand(dst, src2);
1384 }
1385 
1386 void Assembler::bsfl(Register dst, Register src) {
1387   int encode = prefix_and_encode(dst->encoding(), src->encoding());
1388   emit_int8(0x0F);
1389   emit_int8((unsigned char)0xBC);
1390   emit_int8((unsigned char)(0xC0 | encode));
1391 }
1392 
1393 void Assembler::bsrl(Register dst, Register src) {
1394   int encode = prefix_and_encode(dst->encoding(), src->encoding());
1395   emit_int8(0x0F);
1396   emit_int8((unsigned char)0xBD);
1397   emit_int8((unsigned char)(0xC0 | encode));
1398 }
1399 
1400 void Assembler::bswapl(Register reg) { // bswap
1401   int encode = prefix_and_encode(reg->encoding());
1402   emit_int8(0x0F);
1403   emit_int8((unsigned char)(0xC8 | encode));
1404 }
1405 
1406 void Assembler::blsil(Register dst, Register src) {
1407   assert(VM_Version::supports_bmi1(), "bit manipulation instructions not supported");
1408   InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
1409   int encode = vex_prefix_and_encode(rbx->encoding(), dst->encoding(), src->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
1410   emit_int8((unsigned char)0xF3);
1411   emit_int8((unsigned char)(0xC0 | encode));
1412 }
1413 
1414 void Assembler::blsil(Register dst, Address src) {
1415   assert(VM_Version::supports_bmi1(), "bit manipulation instructions not supported");
1416   InstructionMark im(this);
1417   InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
1418   vex_prefix(src, dst->encoding(), rbx->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
1419   emit_int8((unsigned char)0xF3);
1420   emit_operand(rbx, src);
1421 }
1422 
1423 void Assembler::blsmskl(Register dst, Register src) {
1424   assert(VM_Version::supports_bmi1(), "bit manipulation instructions not supported");
1425   InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
1426   int encode = vex_prefix_and_encode(rdx->encoding(), dst->encoding(), src->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
1427   emit_int8((unsigned char)0xF3);
1428   emit_int8((unsigned char)(0xC0 | encode));
1429 }
1430 
1431 void Assembler::blsmskl(Register dst, Address src) {
1432   assert(VM_Version::supports_bmi1(), "bit manipulation instructions not supported");
1433   InstructionMark im(this);
1434   InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
1435   vex_prefix(src, dst->encoding(), rdx->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
1436   emit_int8((unsigned char)0xF3);
1437   emit_operand(rdx, src);
1438 }
1439 
1440 void Assembler::blsrl(Register dst, Register src) {
1441   assert(VM_Version::supports_bmi1(), "bit manipulation instructions not supported");
1442   InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
1443   int encode = vex_prefix_and_encode(rcx->encoding(), dst->encoding(), src->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
1444   emit_int8((unsigned char)0xF3);
1445   emit_int8((unsigned char)(0xC0 | encode));
1446 }
1447 
1448 void Assembler::blsrl(Register dst, Address src) {
1449   assert(VM_Version::supports_bmi1(), "bit manipulation instructions not supported");
1450   InstructionMark im(this);
1451   InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
1452   vex_prefix(src, dst->encoding(), rcx->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
1453   emit_int8((unsigned char)0xF3);
1454   emit_operand(rcx, src);
1455 }
1456 
1457 void Assembler::call(Label& L, relocInfo::relocType rtype) {
1458   // suspect disp32 is always good
1459   int operand = LP64_ONLY(disp32_operand) NOT_LP64(imm_operand);
1460 
1461   if (L.is_bound()) {
1462     const int long_size = 5;
1463     int offs = (int)( target(L) - pc() );
1464     assert(offs <= 0, "assembler error");
1465     InstructionMark im(this);
1466     // 1110 1000 #32-bit disp
1467     emit_int8((unsigned char)0xE8);
1468     emit_data(offs - long_size, rtype, operand);
1469   } else {
1470     InstructionMark im(this);
1471     // 1110 1000 #32-bit disp
1472     L.add_patch_at(code(), locator());
1473 
1474     emit_int8((unsigned char)0xE8);
1475     emit_data(int(0), rtype, operand);
1476   }
1477 }
1478 
1479 void Assembler::call(Register dst) {
1480   int encode = prefix_and_encode(dst->encoding());
1481   emit_int8((unsigned char)0xFF);
1482   emit_int8((unsigned char)(0xD0 | encode));
1483 }
1484 
1485 
1486 void Assembler::call(Address adr) {
1487   InstructionMark im(this);
1488   prefix(adr);
1489   emit_int8((unsigned char)0xFF);
1490   emit_operand(rdx, adr);
1491 }
1492 
1493 void Assembler::call_literal(address entry, RelocationHolder const& rspec) {
1494   assert(entry != NULL, "call most probably wrong");
1495   InstructionMark im(this);
1496   emit_int8((unsigned char)0xE8);
1497   intptr_t disp = entry - (pc() + sizeof(int32_t));
1498   assert(is_simm32(disp), "must be 32bit offset (call2)");
1499   // Technically, should use call32_operand, but this format is
1500   // implied by the fact that we're emitting a call instruction.
1501 
1502   int operand = LP64_ONLY(disp32_operand) NOT_LP64(call32_operand);
1503   emit_data((int) disp, rspec, operand);
1504 }
1505 
1506 void Assembler::cdql() {
1507   emit_int8((unsigned char)0x99);
1508 }
1509 
1510 void Assembler::cld() {
1511   emit_int8((unsigned char)0xFC);
1512 }
1513 
1514 void Assembler::cmovl(Condition cc, Register dst, Register src) {
1515   NOT_LP64(guarantee(VM_Version::supports_cmov(), "illegal instruction"));
1516   int encode = prefix_and_encode(dst->encoding(), src->encoding());
1517   emit_int8(0x0F);
1518   emit_int8(0x40 | cc);
1519   emit_int8((unsigned char)(0xC0 | encode));
1520 }
1521 
1522 
1523 void Assembler::cmovl(Condition cc, Register dst, Address src) {
1524   NOT_LP64(guarantee(VM_Version::supports_cmov(), "illegal instruction"));
1525   prefix(src, dst);
1526   emit_int8(0x0F);
1527   emit_int8(0x40 | cc);
1528   emit_operand(dst, src);
1529 }
1530 
1531 void Assembler::cmpb(Address dst, int imm8) {
1532   InstructionMark im(this);
1533   prefix(dst);
1534   emit_int8((unsigned char)0x80);
1535   emit_operand(rdi, dst, 1);
1536   emit_int8(imm8);
1537 }
1538 
1539 void Assembler::cmpl(Address dst, int32_t imm32) {
1540   InstructionMark im(this);
1541   prefix(dst);
1542   emit_int8((unsigned char)0x81);
1543   emit_operand(rdi, dst, 4);
1544   emit_int32(imm32);
1545 }
1546 
1547 void Assembler::cmpl(Register dst, int32_t imm32) {
1548   prefix(dst);
1549   emit_arith(0x81, 0xF8, dst, imm32);
1550 }
1551 
1552 void Assembler::cmpl(Register dst, Register src) {
1553   (void) prefix_and_encode(dst->encoding(), src->encoding());
1554   emit_arith(0x3B, 0xC0, dst, src);
1555 }
1556 
1557 void Assembler::cmpl(Register dst, Address  src) {
1558   InstructionMark im(this);
1559   prefix(src, dst);
1560   emit_int8((unsigned char)0x3B);
1561   emit_operand(dst, src);
1562 }
1563 
1564 void Assembler::cmpw(Address dst, int imm16) {
1565   InstructionMark im(this);
1566   assert(!dst.base_needs_rex() && !dst.index_needs_rex(), "no extended registers");
1567   emit_int8(0x66);
1568   emit_int8((unsigned char)0x81);
1569   emit_operand(rdi, dst, 2);
1570   emit_int16(imm16);
1571 }
1572 
1573 // The 32-bit cmpxchg compares the value at adr with the contents of rax,
1574 // and stores reg into adr if so; otherwise, the value at adr is loaded into rax,.
1575 // The ZF is set if the compared values were equal, and cleared otherwise.
1576 void Assembler::cmpxchgl(Register reg, Address adr) { // cmpxchg
1577   InstructionMark im(this);
1578   prefix(adr, reg);
1579   emit_int8(0x0F);
1580   emit_int8((unsigned char)0xB1);
1581   emit_operand(reg, adr);
1582 }
1583 
1584 // The 8-bit cmpxchg compares the value at adr with the contents of rax,
1585 // and stores reg into adr if so; otherwise, the value at adr is loaded into rax,.
1586 // The ZF is set if the compared values were equal, and cleared otherwise.
1587 void Assembler::cmpxchgb(Register reg, Address adr) { // cmpxchg
1588   InstructionMark im(this);
1589   prefix(adr, reg, true);
1590   emit_int8(0x0F);
1591   emit_int8((unsigned char)0xB0);
1592   emit_operand(reg, adr);
1593 }
1594 
1595 void Assembler::comisd(XMMRegister dst, Address src) {
1596   // NOTE: dbx seems to decode this as comiss even though the
1597   // 0x66 is there. Strangly ucomisd comes out correct
1598   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
1599   InstructionMark im(this);
1600   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);;
1601   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
1602   simd_prefix(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
1603   emit_int8(0x2F);
1604   emit_operand(dst, src);
1605 }
1606 
1607 void Assembler::comisd(XMMRegister dst, XMMRegister src) {
1608   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
1609   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
1610   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
1611   emit_int8(0x2F);
1612   emit_int8((unsigned char)(0xC0 | encode));
1613 }
1614 
1615 void Assembler::comiss(XMMRegister dst, Address src) {
1616   NOT_LP64(assert(VM_Version::supports_sse(), ""));
1617   InstructionMark im(this);
1618   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
1619   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
1620   simd_prefix(dst, xnoreg, src, VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
1621   emit_int8(0x2F);
1622   emit_operand(dst, src);
1623 }
1624 
1625 void Assembler::comiss(XMMRegister dst, XMMRegister src) {
1626   NOT_LP64(assert(VM_Version::supports_sse(), ""));
1627   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
1628   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
1629   emit_int8(0x2F);
1630   emit_int8((unsigned char)(0xC0 | encode));
1631 }
1632 
1633 void Assembler::cpuid() {
1634   emit_int8(0x0F);
1635   emit_int8((unsigned char)0xA2);
1636 }
1637 
1638 // Opcode / Instruction                      Op /  En  64 - Bit Mode     Compat / Leg Mode Description                  Implemented
1639 // F2 0F 38 F0 / r       CRC32 r32, r / m8   RM        Valid             Valid             Accumulate CRC32 on r / m8.  v
1640 // F2 REX 0F 38 F0 / r   CRC32 r32, r / m8*  RM        Valid             N.E.              Accumulate CRC32 on r / m8.  -
1641 // F2 REX.W 0F 38 F0 / r CRC32 r64, r / m8   RM        Valid             N.E.              Accumulate CRC32 on r / m8.  -
1642 //
1643 // F2 0F 38 F1 / r       CRC32 r32, r / m16  RM        Valid             Valid             Accumulate CRC32 on r / m16. v
1644 //
1645 // F2 0F 38 F1 / r       CRC32 r32, r / m32  RM        Valid             Valid             Accumulate CRC32 on r / m32. v
1646 //
1647 // F2 REX.W 0F 38 F1 / r CRC32 r64, r / m64  RM        Valid             N.E.              Accumulate CRC32 on r / m64. v
1648 void Assembler::crc32(Register crc, Register v, int8_t sizeInBytes) {
1649   assert(VM_Version::supports_sse4_2(), "");
1650   int8_t w = 0x01;
1651   Prefix p = Prefix_EMPTY;
1652 
1653   emit_int8((int8_t)0xF2);
1654   switch (sizeInBytes) {
1655   case 1:
1656     w = 0;
1657     break;
1658   case 2:
1659   case 4:
1660     break;
1661   LP64_ONLY(case 8:)
1662     // This instruction is not valid in 32 bits
1663     // Note:
1664     // http://www.intel.com/content/dam/www/public/us/en/documents/manuals/64-ia-32-architectures-software-developer-instruction-set-reference-manual-325383.pdf
1665     //
1666     // Page B - 72   Vol. 2C says
1667     // qwreg2 to qwreg            1111 0010 : 0100 1R0B : 0000 1111 : 0011 1000 : 1111 0000 : 11 qwreg1 qwreg2
1668     // mem64 to qwreg             1111 0010 : 0100 1R0B : 0000 1111 : 0011 1000 : 1111 0000 : mod qwreg r / m
1669     //                                                                            F0!!!
1670     // while 3 - 208 Vol. 2A
1671     // F2 REX.W 0F 38 F1 / r       CRC32 r64, r / m64             RM         Valid      N.E.Accumulate CRC32 on r / m64.
1672     //
1673     // the 0 on a last bit is reserved for a different flavor of this instruction :
1674     // F2 REX.W 0F 38 F0 / r       CRC32 r64, r / m8              RM         Valid      N.E.Accumulate CRC32 on r / m8.
1675     p = REX_W;
1676     break;
1677   default:
1678     assert(0, "Unsupported value for a sizeInBytes argument");
1679     break;
1680   }
1681   LP64_ONLY(prefix(crc, v, p);)
1682   emit_int8((int8_t)0x0F);
1683   emit_int8(0x38);
1684   emit_int8((int8_t)(0xF0 | w));
1685   emit_int8(0xC0 | ((crc->encoding() & 0x7) << 3) | (v->encoding() & 7));
1686 }
1687 
1688 void Assembler::crc32(Register crc, Address adr, int8_t sizeInBytes) {
1689   assert(VM_Version::supports_sse4_2(), "");
1690   InstructionMark im(this);
1691   int8_t w = 0x01;
1692   Prefix p = Prefix_EMPTY;
1693 
1694   emit_int8((int8_t)0xF2);
1695   switch (sizeInBytes) {
1696   case 1:
1697     w = 0;
1698     break;
1699   case 2:
1700   case 4:
1701     break;
1702   LP64_ONLY(case 8:)
1703     // This instruction is not valid in 32 bits
1704     p = REX_W;
1705     break;
1706   default:
1707     assert(0, "Unsupported value for a sizeInBytes argument");
1708     break;
1709   }
1710   LP64_ONLY(prefix(crc, adr, p);)
1711   emit_int8((int8_t)0x0F);
1712   emit_int8(0x38);
1713   emit_int8((int8_t)(0xF0 | w));
1714   emit_operand(crc, adr);
1715 }
1716 
1717 void Assembler::cvtdq2pd(XMMRegister dst, XMMRegister src) {
1718   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
1719   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
1720   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
1721   emit_int8((unsigned char)0xE6);
1722   emit_int8((unsigned char)(0xC0 | encode));
1723 }
1724 
1725 void Assembler::cvtdq2ps(XMMRegister dst, XMMRegister src) {
1726   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
1727   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
1728   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
1729   emit_int8(0x5B);
1730   emit_int8((unsigned char)(0xC0 | encode));
1731 }
1732 
1733 void Assembler::cvtsd2ss(XMMRegister dst, XMMRegister src) {
1734   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
1735   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
1736   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
1737   emit_int8(0x5A);
1738   emit_int8((unsigned char)(0xC0 | encode));
1739 }
1740 
1741 void Assembler::cvtsd2ss(XMMRegister dst, Address src) {
1742   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
1743   InstructionMark im(this);
1744   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
1745   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
1746   simd_prefix(dst, dst, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
1747   emit_int8(0x5A);
1748   emit_operand(dst, src);
1749 }
1750 
1751 void Assembler::cvtsi2sdl(XMMRegister dst, Register src) {
1752   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
1753   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
1754   int encode = simd_prefix_and_encode(dst, dst, as_XMMRegister(src->encoding()), VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
1755   emit_int8(0x2A);
1756   emit_int8((unsigned char)(0xC0 | encode));
1757 }
1758 
1759 void Assembler::cvtsi2sdl(XMMRegister dst, Address src) {
1760   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
1761   InstructionMark im(this);
1762   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
1763   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
1764   simd_prefix(dst, dst, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
1765   emit_int8(0x2A);
1766   emit_operand(dst, src);
1767 }
1768 
1769 void Assembler::cvtsi2ssl(XMMRegister dst, Register src) {
1770   NOT_LP64(assert(VM_Version::supports_sse(), ""));
1771   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
1772   int encode = simd_prefix_and_encode(dst, dst, as_XMMRegister(src->encoding()), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
1773   emit_int8(0x2A);
1774   emit_int8((unsigned char)(0xC0 | encode));
1775 }
1776 
1777 void Assembler::cvtsi2ssl(XMMRegister dst, Address src) {
1778   NOT_LP64(assert(VM_Version::supports_sse(), ""));
1779   InstructionMark im(this);
1780   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
1781   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
1782   simd_prefix(dst, dst, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
1783   emit_int8(0x2A);
1784   emit_operand(dst, src);
1785 }
1786 
1787 void Assembler::cvtsi2ssq(XMMRegister dst, Register src) {
1788   NOT_LP64(assert(VM_Version::supports_sse(), ""));
1789   InstructionAttr attributes(AVX_128bit, /* rex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
1790   int encode = simd_prefix_and_encode(dst, dst, as_XMMRegister(src->encoding()), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
1791   emit_int8(0x2A);
1792   emit_int8((unsigned char)(0xC0 | encode));
1793 }
1794 
1795 void Assembler::cvtss2sd(XMMRegister dst, XMMRegister src) {
1796   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
1797   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
1798   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
1799   emit_int8(0x5A);
1800   emit_int8((unsigned char)(0xC0 | encode));
1801 }
1802 
1803 void Assembler::cvtss2sd(XMMRegister dst, Address src) {
1804   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
1805   InstructionMark im(this);
1806   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
1807   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
1808   simd_prefix(dst, dst, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
1809   emit_int8(0x5A);
1810   emit_operand(dst, src);
1811 }
1812 
1813 
1814 void Assembler::cvttsd2sil(Register dst, XMMRegister src) {
1815   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
1816   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
1817   int encode = simd_prefix_and_encode(as_XMMRegister(dst->encoding()), xnoreg, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
1818   emit_int8(0x2C);
1819   emit_int8((unsigned char)(0xC0 | encode));
1820 }
1821 
1822 void Assembler::cvttss2sil(Register dst, XMMRegister src) {
1823   NOT_LP64(assert(VM_Version::supports_sse(), ""));
1824   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
1825   int encode = simd_prefix_and_encode(as_XMMRegister(dst->encoding()), xnoreg, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
1826   emit_int8(0x2C);
1827   emit_int8((unsigned char)(0xC0 | encode));
1828 }
1829 
1830 void Assembler::cvttpd2dq(XMMRegister dst, XMMRegister src) {
1831   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
1832   int vector_len = VM_Version::supports_avx512novl() ? AVX_512bit : AVX_128bit;
1833   InstructionAttr attributes(vector_len, /* rex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
1834   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
1835   emit_int8((unsigned char)0xE6);
1836   emit_int8((unsigned char)(0xC0 | encode));
1837 }
1838 
1839 void Assembler::decl(Address dst) {
1840   // Don't use it directly. Use MacroAssembler::decrement() instead.
1841   InstructionMark im(this);
1842   prefix(dst);
1843   emit_int8((unsigned char)0xFF);
1844   emit_operand(rcx, dst);
1845 }
1846 
1847 void Assembler::divsd(XMMRegister dst, Address src) {
1848   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
1849   InstructionMark im(this);
1850   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
1851   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
1852   simd_prefix(dst, dst, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
1853   emit_int8(0x5E);
1854   emit_operand(dst, src);
1855 }
1856 
1857 void Assembler::divsd(XMMRegister dst, XMMRegister src) {
1858   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
1859   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
1860   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
1861   emit_int8(0x5E);
1862   emit_int8((unsigned char)(0xC0 | encode));
1863 }
1864 
1865 void Assembler::divss(XMMRegister dst, Address src) {
1866   NOT_LP64(assert(VM_Version::supports_sse(), ""));
1867   InstructionMark im(this);
1868   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
1869   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
1870   simd_prefix(dst, dst, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
1871   emit_int8(0x5E);
1872   emit_operand(dst, src);
1873 }
1874 
1875 void Assembler::divss(XMMRegister dst, XMMRegister src) {
1876   NOT_LP64(assert(VM_Version::supports_sse(), ""));
1877   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
1878   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
1879   emit_int8(0x5E);
1880   emit_int8((unsigned char)(0xC0 | encode));
1881 }
1882 
1883 void Assembler::emms() {
1884   NOT_LP64(assert(VM_Version::supports_mmx(), ""));
1885   emit_int8(0x0F);
1886   emit_int8(0x77);
1887 }
1888 
1889 void Assembler::hlt() {
1890   emit_int8((unsigned char)0xF4);
1891 }
1892 
1893 void Assembler::idivl(Register src) {
1894   int encode = prefix_and_encode(src->encoding());
1895   emit_int8((unsigned char)0xF7);
1896   emit_int8((unsigned char)(0xF8 | encode));
1897 }
1898 
1899 void Assembler::divl(Register src) { // Unsigned
1900   int encode = prefix_and_encode(src->encoding());
1901   emit_int8((unsigned char)0xF7);
1902   emit_int8((unsigned char)(0xF0 | encode));
1903 }
1904 
1905 void Assembler::imull(Register src) {
1906   int encode = prefix_and_encode(src->encoding());
1907   emit_int8((unsigned char)0xF7);
1908   emit_int8((unsigned char)(0xE8 | encode));
1909 }
1910 
1911 void Assembler::imull(Register dst, Register src) {
1912   int encode = prefix_and_encode(dst->encoding(), src->encoding());
1913   emit_int8(0x0F);
1914   emit_int8((unsigned char)0xAF);
1915   emit_int8((unsigned char)(0xC0 | encode));
1916 }
1917 
1918 
1919 void Assembler::imull(Register dst, Register src, int value) {
1920   int encode = prefix_and_encode(dst->encoding(), src->encoding());
1921   if (is8bit(value)) {
1922     emit_int8(0x6B);
1923     emit_int8((unsigned char)(0xC0 | encode));
1924     emit_int8(value & 0xFF);
1925   } else {
1926     emit_int8(0x69);
1927     emit_int8((unsigned char)(0xC0 | encode));
1928     emit_int32(value);
1929   }
1930 }
1931 
1932 void Assembler::imull(Register dst, Address src) {
1933   InstructionMark im(this);
1934   prefix(src, dst);
1935   emit_int8(0x0F);
1936   emit_int8((unsigned char) 0xAF);
1937   emit_operand(dst, src);
1938 }
1939 
1940 
1941 void Assembler::incl(Address dst) {
1942   // Don't use it directly. Use MacroAssembler::increment() instead.
1943   InstructionMark im(this);
1944   prefix(dst);
1945   emit_int8((unsigned char)0xFF);
1946   emit_operand(rax, dst);
1947 }
1948 
1949 void Assembler::jcc(Condition cc, Label& L, bool maybe_short) {
1950   InstructionMark im(this);
1951   assert((0 <= cc) && (cc < 16), "illegal cc");
1952   if (L.is_bound()) {
1953     address dst = target(L);
1954     assert(dst != NULL, "jcc most probably wrong");
1955 
1956     const int short_size = 2;
1957     const int long_size = 6;
1958     intptr_t offs = (intptr_t)dst - (intptr_t)pc();
1959     if (maybe_short && is8bit(offs - short_size)) {
1960       // 0111 tttn #8-bit disp
1961       emit_int8(0x70 | cc);
1962       emit_int8((offs - short_size) & 0xFF);
1963     } else {
1964       // 0000 1111 1000 tttn #32-bit disp
1965       assert(is_simm32(offs - long_size),
1966              "must be 32bit offset (call4)");
1967       emit_int8(0x0F);
1968       emit_int8((unsigned char)(0x80 | cc));
1969       emit_int32(offs - long_size);
1970     }
1971   } else {
1972     // Note: could eliminate cond. jumps to this jump if condition
1973     //       is the same however, seems to be rather unlikely case.
1974     // Note: use jccb() if label to be bound is very close to get
1975     //       an 8-bit displacement
1976     L.add_patch_at(code(), locator());
1977     emit_int8(0x0F);
1978     emit_int8((unsigned char)(0x80 | cc));
1979     emit_int32(0);
1980   }
1981 }
1982 
1983 void Assembler::jccb(Condition cc, Label& L) {
1984   if (L.is_bound()) {
1985     const int short_size = 2;
1986     address entry = target(L);
1987 #ifdef ASSERT
1988     intptr_t dist = (intptr_t)entry - ((intptr_t)pc() + short_size);
1989     intptr_t delta = short_branch_delta();
1990     if (delta != 0) {
1991       dist += (dist < 0 ? (-delta) :delta);
1992     }
1993     assert(is8bit(dist), "Dispacement too large for a short jmp");
1994 #endif
1995     intptr_t offs = (intptr_t)entry - (intptr_t)pc();
1996     // 0111 tttn #8-bit disp
1997     emit_int8(0x70 | cc);
1998     emit_int8((offs - short_size) & 0xFF);
1999   } else {
2000     InstructionMark im(this);
2001     L.add_patch_at(code(), locator());
2002     emit_int8(0x70 | cc);
2003     emit_int8(0);
2004   }
2005 }
2006 
2007 void Assembler::jmp(Address adr) {
2008   InstructionMark im(this);
2009   prefix(adr);
2010   emit_int8((unsigned char)0xFF);
2011   emit_operand(rsp, adr);
2012 }
2013 
2014 void Assembler::jmp(Label& L, bool maybe_short) {
2015   if (L.is_bound()) {
2016     address entry = target(L);
2017     assert(entry != NULL, "jmp most probably wrong");
2018     InstructionMark im(this);
2019     const int short_size = 2;
2020     const int long_size = 5;
2021     intptr_t offs = entry - pc();
2022     if (maybe_short && is8bit(offs - short_size)) {
2023       emit_int8((unsigned char)0xEB);
2024       emit_int8((offs - short_size) & 0xFF);
2025     } else {
2026       emit_int8((unsigned char)0xE9);
2027       emit_int32(offs - long_size);
2028     }
2029   } else {
2030     // By default, forward jumps are always 32-bit displacements, since
2031     // we can't yet know where the label will be bound.  If you're sure that
2032     // the forward jump will not run beyond 256 bytes, use jmpb to
2033     // force an 8-bit displacement.
2034     InstructionMark im(this);
2035     L.add_patch_at(code(), locator());
2036     emit_int8((unsigned char)0xE9);
2037     emit_int32(0);
2038   }
2039 }
2040 
2041 void Assembler::jmp(Register entry) {
2042   int encode = prefix_and_encode(entry->encoding());
2043   emit_int8((unsigned char)0xFF);
2044   emit_int8((unsigned char)(0xE0 | encode));
2045 }
2046 
2047 void Assembler::jmp_literal(address dest, RelocationHolder const& rspec) {
2048   InstructionMark im(this);
2049   emit_int8((unsigned char)0xE9);
2050   assert(dest != NULL, "must have a target");
2051   intptr_t disp = dest - (pc() + sizeof(int32_t));
2052   assert(is_simm32(disp), "must be 32bit offset (jmp)");
2053   emit_data(disp, rspec.reloc(), call32_operand);
2054 }
2055 
2056 void Assembler::jmpb(Label& L) {
2057   if (L.is_bound()) {
2058     const int short_size = 2;
2059     address entry = target(L);
2060     assert(entry != NULL, "jmp most probably wrong");
2061 #ifdef ASSERT
2062     intptr_t dist = (intptr_t)entry - ((intptr_t)pc() + short_size);
2063     intptr_t delta = short_branch_delta();
2064     if (delta != 0) {
2065       dist += (dist < 0 ? (-delta) :delta);
2066     }
2067     assert(is8bit(dist), "Dispacement too large for a short jmp");
2068 #endif
2069     intptr_t offs = entry - pc();
2070     emit_int8((unsigned char)0xEB);
2071     emit_int8((offs - short_size) & 0xFF);
2072   } else {
2073     InstructionMark im(this);
2074     L.add_patch_at(code(), locator());
2075     emit_int8((unsigned char)0xEB);
2076     emit_int8(0);
2077   }
2078 }
2079 
2080 void Assembler::ldmxcsr( Address src) {
2081   NOT_LP64(assert(VM_Version::supports_sse(), ""));
2082   InstructionMark im(this);
2083   prefix(src);
2084   emit_int8(0x0F);
2085   emit_int8((unsigned char)0xAE);
2086   emit_operand(as_Register(2), src);
2087 }
2088 
2089 void Assembler::leal(Register dst, Address src) {
2090   InstructionMark im(this);
2091 #ifdef _LP64
2092   emit_int8(0x67); // addr32
2093   prefix(src, dst);
2094 #endif // LP64
2095   emit_int8((unsigned char)0x8D);
2096   emit_operand(dst, src);
2097 }
2098 
2099 void Assembler::lfence() {
2100   emit_int8(0x0F);
2101   emit_int8((unsigned char)0xAE);
2102   emit_int8((unsigned char)0xE8);
2103 }
2104 
2105 void Assembler::lock() {
2106   emit_int8((unsigned char)0xF0);
2107 }
2108 
2109 void Assembler::lzcntl(Register dst, Register src) {
2110   assert(VM_Version::supports_lzcnt(), "encoding is treated as BSR");
2111   emit_int8((unsigned char)0xF3);
2112   int encode = prefix_and_encode(dst->encoding(), src->encoding());
2113   emit_int8(0x0F);
2114   emit_int8((unsigned char)0xBD);
2115   emit_int8((unsigned char)(0xC0 | encode));
2116 }
2117 
2118 // Emit mfence instruction
2119 void Assembler::mfence() {
2120   NOT_LP64(assert(VM_Version::supports_sse2(), "unsupported");)
2121   emit_int8(0x0F);
2122   emit_int8((unsigned char)0xAE);
2123   emit_int8((unsigned char)0xF0);
2124 }
2125 
2126 void Assembler::mov(Register dst, Register src) {
2127   LP64_ONLY(movq(dst, src)) NOT_LP64(movl(dst, src));
2128 }
2129 
2130 void Assembler::movapd(XMMRegister dst, XMMRegister src) {
2131   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
2132   int vector_len = VM_Version::supports_avx512novl() ? AVX_512bit : AVX_128bit;
2133   InstructionAttr attributes(vector_len, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
2134   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
2135   emit_int8(0x28);
2136   emit_int8((unsigned char)(0xC0 | encode));
2137 }
2138 
2139 void Assembler::movaps(XMMRegister dst, XMMRegister src) {
2140   NOT_LP64(assert(VM_Version::supports_sse(), ""));
2141   int vector_len = VM_Version::supports_avx512novl() ? AVX_512bit : AVX_128bit;
2142   InstructionAttr attributes(vector_len, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
2143   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
2144   emit_int8(0x28);
2145   emit_int8((unsigned char)(0xC0 | encode));
2146 }
2147 
2148 void Assembler::movlhps(XMMRegister dst, XMMRegister src) {
2149   NOT_LP64(assert(VM_Version::supports_sse(), ""));
2150   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
2151   int encode = simd_prefix_and_encode(dst, src, src, VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
2152   emit_int8(0x16);
2153   emit_int8((unsigned char)(0xC0 | encode));
2154 }
2155 
2156 void Assembler::movb(Register dst, Address src) {
2157   NOT_LP64(assert(dst->has_byte_register(), "must have byte register"));
2158   InstructionMark im(this);
2159   prefix(src, dst, true);
2160   emit_int8((unsigned char)0x8A);
2161   emit_operand(dst, src);
2162 }
2163 
2164 void Assembler::movddup(XMMRegister dst, XMMRegister src) {
2165   NOT_LP64(assert(VM_Version::supports_sse3(), ""));
2166   int vector_len = VM_Version::supports_avx512novl() ? AVX_512bit : AVX_128bit;
2167   InstructionAttr attributes(vector_len, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
2168   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
2169   emit_int8(0x12);
2170   emit_int8(0xC0 | encode);
2171 }
2172 
2173 void Assembler::kmovbl(KRegister dst, Register src) {
2174   assert(VM_Version::supports_avx512dq(), "");
2175   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ true, /* uses_vl */ false);
2176   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
2177   emit_int8((unsigned char)0x92);
2178   emit_int8((unsigned char)(0xC0 | encode));
2179 }
2180 
2181 void Assembler::kmovbl(Register dst, KRegister src) {
2182   assert(VM_Version::supports_avx512dq(), "");
2183   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ true, /* uses_vl */ false);
2184   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
2185   emit_int8((unsigned char)0x93);
2186   emit_int8((unsigned char)(0xC0 | encode));
2187 }
2188 
2189 void Assembler::kmovwl(KRegister dst, Register src) {
2190   assert(VM_Version::supports_evex(), "");
2191   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ true, /* uses_vl */ false);
2192   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
2193   emit_int8((unsigned char)0x92);
2194   emit_int8((unsigned char)(0xC0 | encode));
2195 }
2196 
2197 void Assembler::kmovwl(Register dst, KRegister src) {
2198   assert(VM_Version::supports_evex(), "");
2199   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ true, /* uses_vl */ false);
2200   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
2201   emit_int8((unsigned char)0x93);
2202   emit_int8((unsigned char)(0xC0 | encode));
2203 }
2204 
2205 void Assembler::kmovdl(KRegister dst, Register src) {
2206   assert(VM_Version::supports_avx512bw(), "");
2207   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ true, /* uses_vl */ false);
2208   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
2209   emit_int8((unsigned char)0x92);
2210   emit_int8((unsigned char)(0xC0 | encode));
2211 }
2212 
2213 void Assembler::kmovdl(Register dst, KRegister src) {
2214   assert(VM_Version::supports_avx512bw(), "");
2215   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ true, /* uses_vl */ false);
2216   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
2217   emit_int8((unsigned char)0x93);
2218   emit_int8((unsigned char)(0xC0 | encode));
2219 }
2220 
2221 void Assembler::kmovql(KRegister dst, KRegister src) {
2222   assert(VM_Version::supports_avx512bw(), "");
2223   InstructionAttr attributes(AVX_128bit, /* rex_w */ true, /* legacy_mode */ true, /* no_mask_reg */ true, /* uses_vl */ false);
2224   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
2225   emit_int8((unsigned char)0x90);
2226   emit_int8((unsigned char)(0xC0 | encode));
2227 }
2228 
2229 void Assembler::kmovql(KRegister dst, Address src) {
2230   assert(VM_Version::supports_avx512bw(), "");
2231   InstructionMark im(this);
2232   InstructionAttr attributes(AVX_128bit, /* vex_w */ true, /* legacy_mode */ true, /* no_mask_reg */ true, /* uses_vl */ false);
2233   vex_prefix(src, 0, dst->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
2234   emit_int8((unsigned char)0x90);
2235   emit_operand((Register)dst, src);
2236 }
2237 
2238 void Assembler::kmovql(Address dst, KRegister src) {
2239   assert(VM_Version::supports_avx512bw(), "");
2240   InstructionMark im(this);
2241   InstructionAttr attributes(AVX_128bit, /* vex_w */ true, /* legacy_mode */ true, /* no_mask_reg */ true, /* uses_vl */ false);
2242   vex_prefix(dst, 0, src->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
2243   emit_int8((unsigned char)0x90);
2244   emit_operand((Register)src, dst);
2245 }
2246 
2247 void Assembler::kmovql(KRegister dst, Register src) {
2248   assert(VM_Version::supports_avx512bw(), "");
2249   InstructionAttr attributes(AVX_128bit, /* rex_w */ true, /* legacy_mode */ true, /* no_mask_reg */ true, /* uses_vl */ false);
2250   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
2251   emit_int8((unsigned char)0x92);
2252   emit_int8((unsigned char)(0xC0 | encode));
2253 }
2254 
2255 void Assembler::kmovql(Register dst, KRegister src) {
2256   assert(VM_Version::supports_avx512bw(), "");
2257   InstructionAttr attributes(AVX_128bit, /* rex_w */ true, /* legacy_mode */ true, /* no_mask_reg */ true, /* uses_vl */ false);
2258   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
2259   emit_int8((unsigned char)0x93);
2260   emit_int8((unsigned char)(0xC0 | encode));
2261 }
2262 
2263 // This instruction produces ZF or CF flags
2264 void Assembler::kortestbl(KRegister src1, KRegister src2) {
2265   assert(VM_Version::supports_avx512dq(), "");
2266   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ true, /* uses_vl */ false);
2267   int encode = vex_prefix_and_encode(src1->encoding(), 0, src2->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
2268   emit_int8((unsigned char)0x98);
2269   emit_int8((unsigned char)(0xC0 | encode));
2270 }
2271 
2272 // This instruction produces ZF or CF flags
2273 void Assembler::kortestwl(KRegister src1, KRegister src2) {
2274   assert(VM_Version::supports_evex(), "");
2275   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ true, /* uses_vl */ false);
2276   int encode = vex_prefix_and_encode(src1->encoding(), 0, src2->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
2277   emit_int8((unsigned char)0x98);
2278   emit_int8((unsigned char)(0xC0 | encode));
2279 }
2280 
2281 // This instruction produces ZF or CF flags
2282 void Assembler::kortestdl(KRegister src1, KRegister src2) {
2283   assert(VM_Version::supports_avx512bw(), "");
2284   InstructionAttr attributes(AVX_128bit, /* rex_w */ true, /* legacy_mode */ true, /* no_mask_reg */ true, /* uses_vl */ false);
2285   int encode = vex_prefix_and_encode(src1->encoding(), 0, src2->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
2286   emit_int8((unsigned char)0x98);
2287   emit_int8((unsigned char)(0xC0 | encode));
2288 }
2289 
2290 // This instruction produces ZF or CF flags
2291 void Assembler::kortestql(KRegister src1, KRegister src2) {
2292   assert(VM_Version::supports_avx512bw(), "");
2293   InstructionAttr attributes(AVX_128bit, /* rex_w */ true, /* legacy_mode */ true, /* no_mask_reg */ true, /* uses_vl */ false);
2294   int encode = vex_prefix_and_encode(src1->encoding(), 0, src2->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
2295   emit_int8((unsigned char)0x98);
2296   emit_int8((unsigned char)(0xC0 | encode));
2297 }
2298 
2299 void Assembler::movb(Address dst, int imm8) {
2300   InstructionMark im(this);
2301    prefix(dst);
2302   emit_int8((unsigned char)0xC6);
2303   emit_operand(rax, dst, 1);
2304   emit_int8(imm8);
2305 }
2306 
2307 
2308 void Assembler::movb(Address dst, Register src) {
2309   assert(src->has_byte_register(), "must have byte register");
2310   InstructionMark im(this);
2311   prefix(dst, src, true);
2312   emit_int8((unsigned char)0x88);
2313   emit_operand(src, dst);
2314 }
2315 
2316 void Assembler::movdl(XMMRegister dst, Register src) {
2317   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
2318   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
2319   int encode = simd_prefix_and_encode(dst, xnoreg, as_XMMRegister(src->encoding()), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
2320   emit_int8(0x6E);
2321   emit_int8((unsigned char)(0xC0 | encode));
2322 }
2323 
2324 void Assembler::movdl(Register dst, XMMRegister src) {
2325   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
2326   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
2327   // swap src/dst to get correct prefix
2328   int encode = simd_prefix_and_encode(src, xnoreg, as_XMMRegister(dst->encoding()), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
2329   emit_int8(0x7E);
2330   emit_int8((unsigned char)(0xC0 | encode));
2331 }
2332 
2333 void Assembler::movdl(XMMRegister dst, Address src) {
2334   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
2335   InstructionMark im(this);
2336   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
2337   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
2338   simd_prefix(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
2339   emit_int8(0x6E);
2340   emit_operand(dst, src);
2341 }
2342 
2343 void Assembler::movdl(Address dst, XMMRegister src) {
2344   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
2345   InstructionMark im(this);
2346   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
2347   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
2348   simd_prefix(src, xnoreg, dst, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
2349   emit_int8(0x7E);
2350   emit_operand(src, dst);
2351 }
2352 
2353 void Assembler::movdqa(XMMRegister dst, XMMRegister src) {
2354   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
2355   int vector_len = VM_Version::supports_avx512novl() ? AVX_512bit : AVX_128bit;
2356   InstructionAttr attributes(vector_len, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
2357   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
2358   emit_int8(0x6F);
2359   emit_int8((unsigned char)(0xC0 | encode));
2360 }
2361 
2362 void Assembler::movdqa(XMMRegister dst, Address src) {
2363   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
2364   InstructionMark im(this);
2365   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
2366   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
2367   simd_prefix(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
2368   emit_int8(0x6F);
2369   emit_operand(dst, src);
2370 }
2371 
2372 void Assembler::movdqu(XMMRegister dst, Address src) {
2373   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
2374   InstructionMark im(this);
2375   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
2376   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
2377   simd_prefix(dst, xnoreg, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
2378   emit_int8(0x6F);
2379   emit_operand(dst, src);
2380 }
2381 
2382 void Assembler::movdqu(XMMRegister dst, XMMRegister src) {
2383   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
2384   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
2385   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
2386   emit_int8(0x6F);
2387   emit_int8((unsigned char)(0xC0 | encode));
2388 }
2389 
2390 void Assembler::movdqu(Address dst, XMMRegister src) {
2391   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
2392   InstructionMark im(this);
2393   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
2394   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
2395   simd_prefix(src, xnoreg, dst, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
2396   emit_int8(0x7F);
2397   emit_operand(src, dst);
2398 }
2399 
2400 // Move Unaligned 256bit Vector
2401 void Assembler::vmovdqu(XMMRegister dst, XMMRegister src) {
2402   assert(UseAVX > 0, "");
2403   InstructionAttr attributes(AVX_256bit, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
2404   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
2405   emit_int8(0x6F);
2406   emit_int8((unsigned char)(0xC0 | encode));
2407 }
2408 
2409 void Assembler::vmovdqu(XMMRegister dst, Address src) {
2410   assert(UseAVX > 0, "");
2411   InstructionMark im(this);
2412   InstructionAttr attributes(AVX_256bit, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
2413   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
2414   vex_prefix(src, 0, dst->encoding(), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
2415   emit_int8(0x6F);
2416   emit_operand(dst, src);
2417 }
2418 
2419 void Assembler::vmovdqu(Address dst, XMMRegister src) {
2420   assert(UseAVX > 0, "");
2421   InstructionMark im(this);
2422   InstructionAttr attributes(AVX_256bit, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
2423   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
2424   // swap src<->dst for encoding
2425   assert(src != xnoreg, "sanity");
2426   vex_prefix(dst, 0, src->encoding(), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
2427   emit_int8(0x7F);
2428   emit_operand(src, dst);
2429 }
2430 
2431 // Move Unaligned EVEX enabled Vector (programmable : 8,16,32,64)
2432 void Assembler::evmovdqub(XMMRegister dst, XMMRegister src, int vector_len) {
2433   assert(VM_Version::supports_evex(), "");
2434   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
2435   int prefix = (_legacy_mode_bw) ? VEX_SIMD_F2 : VEX_SIMD_F3;
2436   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), (Assembler::VexSimdPrefix)prefix, VEX_OPCODE_0F, &attributes);
2437   emit_int8(0x6F);
2438   emit_int8((unsigned char)(0xC0 | encode));
2439 }
2440 
2441 void Assembler::evmovdqub(XMMRegister dst, Address src, int vector_len) {
2442   assert(VM_Version::supports_evex(), "");
2443   InstructionMark im(this);
2444   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
2445   int prefix = (_legacy_mode_bw) ? VEX_SIMD_F2 : VEX_SIMD_F3;
2446   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
2447   vex_prefix(src, 0, dst->encoding(), (Assembler::VexSimdPrefix)prefix, VEX_OPCODE_0F, &attributes);
2448   emit_int8(0x6F);
2449   emit_operand(dst, src);
2450 }
2451 
2452 void Assembler::evmovdqub(Address dst, XMMRegister src, int vector_len) {
2453   assert(VM_Version::supports_evex(), "");
2454   assert(src != xnoreg, "sanity");
2455   InstructionMark im(this);
2456   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
2457   int prefix = (_legacy_mode_bw) ? VEX_SIMD_F2 : VEX_SIMD_F3;
2458   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
2459   vex_prefix(dst, 0, src->encoding(), (Assembler::VexSimdPrefix)prefix, VEX_OPCODE_0F, &attributes);
2460   emit_int8(0x7F);
2461   emit_operand(src, dst);
2462 }
2463 
2464 void Assembler::evmovdquw(XMMRegister dst, XMMRegister src, int vector_len) {
2465   assert(VM_Version::supports_evex(), "");
2466   InstructionAttr attributes(vector_len, /* vex_w */ true, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
2467   int prefix = (_legacy_mode_bw) ? VEX_SIMD_F2 : VEX_SIMD_F3;
2468   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), (Assembler::VexSimdPrefix)prefix, VEX_OPCODE_0F, &attributes);
2469   emit_int8(0x6F);
2470   emit_int8((unsigned char)(0xC0 | encode));
2471 }
2472 
2473 void Assembler::evmovdquw(XMMRegister dst, Address src, int vector_len) {
2474   assert(VM_Version::supports_evex(), "");
2475   InstructionMark im(this);
2476   InstructionAttr attributes(vector_len, /* vex_w */ true, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
2477   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
2478   int prefix = (_legacy_mode_bw) ? VEX_SIMD_F2 : VEX_SIMD_F3;
2479   vex_prefix(src, 0, dst->encoding(), (Assembler::VexSimdPrefix)prefix, VEX_OPCODE_0F, &attributes);
2480   emit_int8(0x6F);
2481   emit_operand(dst, src);
2482 }
2483 
2484 void Assembler::evmovdquw(Address dst, XMMRegister src, int vector_len) {
2485   assert(VM_Version::supports_evex(), "");
2486   assert(src != xnoreg, "sanity");
2487   InstructionMark im(this);
2488   InstructionAttr attributes(vector_len, /* vex_w */ true, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
2489   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
2490   int prefix = (_legacy_mode_bw) ? VEX_SIMD_F2 : VEX_SIMD_F3;
2491   vex_prefix(dst, 0, src->encoding(), (Assembler::VexSimdPrefix)prefix, VEX_OPCODE_0F, &attributes);
2492   emit_int8(0x7F);
2493   emit_operand(src, dst);
2494 }
2495 
2496 void Assembler::evmovdqul(XMMRegister dst, XMMRegister src, int vector_len) {
2497   assert(VM_Version::supports_evex(), "");
2498   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
2499   attributes.set_is_evex_instruction();
2500   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
2501   emit_int8(0x6F);
2502   emit_int8((unsigned char)(0xC0 | encode));
2503 }
2504 
2505 void Assembler::evmovdqul(XMMRegister dst, Address src, int vector_len) {
2506   assert(VM_Version::supports_evex(), "");
2507   InstructionMark im(this);
2508   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false , /* uses_vl */ true);
2509   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
2510   attributes.set_is_evex_instruction();
2511   vex_prefix(src, 0, dst->encoding(), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
2512   emit_int8(0x6F);
2513   emit_operand(dst, src);
2514 }
2515 
2516 void Assembler::evmovdqul(Address dst, XMMRegister src, int vector_len) {
2517   assert(VM_Version::supports_evex(), "");
2518   assert(src != xnoreg, "sanity");
2519   InstructionMark im(this);
2520   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
2521   attributes.set_is_evex_instruction();
2522   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
2523   vex_prefix(dst, 0, src->encoding(), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
2524   emit_int8(0x7F);
2525   emit_operand(src, dst);
2526 }
2527 
2528 void Assembler::evmovdquq(XMMRegister dst, XMMRegister src, int vector_len) {
2529   assert(VM_Version::supports_evex(), "");
2530   InstructionAttr attributes(vector_len, /* vex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
2531   attributes.set_is_evex_instruction();
2532   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
2533   emit_int8(0x6F);
2534   emit_int8((unsigned char)(0xC0 | encode));
2535 }
2536 
2537 void Assembler::evmovdquq(XMMRegister dst, Address src, int vector_len) {
2538   assert(VM_Version::supports_evex(), "");
2539   InstructionMark im(this);
2540   InstructionAttr attributes(vector_len, /* vex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
2541   attributes.set_is_evex_instruction();
2542   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
2543   vex_prefix(src, 0, dst->encoding(), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
2544   emit_int8(0x6F);
2545   emit_operand(dst, src);
2546 }
2547 
2548 void Assembler::evmovdquq(Address dst, XMMRegister src, int vector_len) {
2549   assert(VM_Version::supports_evex(), "");
2550   assert(src != xnoreg, "sanity");
2551   InstructionMark im(this);
2552   InstructionAttr attributes(vector_len, /* vex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
2553   attributes.set_is_evex_instruction();
2554   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
2555   vex_prefix(dst, 0, src->encoding(), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
2556   emit_int8(0x7F);
2557   emit_operand(src, dst);
2558 }
2559 
2560 // Uses zero extension on 64bit
2561 
2562 void Assembler::movl(Register dst, int32_t imm32) {
2563   int encode = prefix_and_encode(dst->encoding());
2564   emit_int8((unsigned char)(0xB8 | encode));
2565   emit_int32(imm32);
2566 }
2567 
2568 void Assembler::movl(Register dst, Register src) {
2569   int encode = prefix_and_encode(dst->encoding(), src->encoding());
2570   emit_int8((unsigned char)0x8B);
2571   emit_int8((unsigned char)(0xC0 | encode));
2572 }
2573 
2574 void Assembler::movl(Register dst, Address src) {
2575   InstructionMark im(this);
2576   prefix(src, dst);
2577   emit_int8((unsigned char)0x8B);
2578   emit_operand(dst, src);
2579 }
2580 
2581 void Assembler::movl(Address dst, int32_t imm32) {
2582   InstructionMark im(this);
2583   prefix(dst);
2584   emit_int8((unsigned char)0xC7);
2585   emit_operand(rax, dst, 4);
2586   emit_int32(imm32);
2587 }
2588 
2589 void Assembler::movl(Address dst, Register src) {
2590   InstructionMark im(this);
2591   prefix(dst, src);
2592   emit_int8((unsigned char)0x89);
2593   emit_operand(src, dst);
2594 }
2595 
2596 // New cpus require to use movsd and movss to avoid partial register stall
2597 // when loading from memory. But for old Opteron use movlpd instead of movsd.
2598 // The selection is done in MacroAssembler::movdbl() and movflt().
2599 void Assembler::movlpd(XMMRegister dst, Address src) {
2600   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
2601   InstructionMark im(this);
2602   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
2603   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
2604   simd_prefix(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
2605   emit_int8(0x12);
2606   emit_operand(dst, src);
2607 }
2608 
2609 void Assembler::movq( MMXRegister dst, Address src ) {
2610   assert( VM_Version::supports_mmx(), "" );
2611   emit_int8(0x0F);
2612   emit_int8(0x6F);
2613   emit_operand(dst, src);
2614 }
2615 
2616 void Assembler::movq( Address dst, MMXRegister src ) {
2617   assert( VM_Version::supports_mmx(), "" );
2618   emit_int8(0x0F);
2619   emit_int8(0x7F);
2620   // workaround gcc (3.2.1-7a) bug
2621   // In that version of gcc with only an emit_operand(MMX, Address)
2622   // gcc will tail jump and try and reverse the parameters completely
2623   // obliterating dst in the process. By having a version available
2624   // that doesn't need to swap the args at the tail jump the bug is
2625   // avoided.
2626   emit_operand(dst, src);
2627 }
2628 
2629 void Assembler::movq(XMMRegister dst, Address src) {
2630   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
2631   InstructionMark im(this);
2632   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
2633   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
2634   simd_prefix(dst, xnoreg, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
2635   emit_int8(0x7E);
2636   emit_operand(dst, src);
2637 }
2638 
2639 void Assembler::movq(Address dst, XMMRegister src) {
2640   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
2641   InstructionMark im(this);
2642   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
2643   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
2644   simd_prefix(src, xnoreg, dst, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
2645   emit_int8((unsigned char)0xD6);
2646   emit_operand(src, dst);
2647 }
2648 
2649 void Assembler::movsbl(Register dst, Address src) { // movsxb
2650   InstructionMark im(this);
2651   prefix(src, dst);
2652   emit_int8(0x0F);
2653   emit_int8((unsigned char)0xBE);
2654   emit_operand(dst, src);
2655 }
2656 
2657 void Assembler::movsbl(Register dst, Register src) { // movsxb
2658   NOT_LP64(assert(src->has_byte_register(), "must have byte register"));
2659   int encode = prefix_and_encode(dst->encoding(), false, src->encoding(), true);
2660   emit_int8(0x0F);
2661   emit_int8((unsigned char)0xBE);
2662   emit_int8((unsigned char)(0xC0 | encode));
2663 }
2664 
2665 void Assembler::movsd(XMMRegister dst, XMMRegister src) {
2666   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
2667   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
2668   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
2669   emit_int8(0x10);
2670   emit_int8((unsigned char)(0xC0 | encode));
2671 }
2672 
2673 void Assembler::movsd(XMMRegister dst, Address src) {
2674   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
2675   InstructionMark im(this);
2676   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
2677   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
2678   simd_prefix(dst, xnoreg, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
2679   emit_int8(0x10);
2680   emit_operand(dst, src);
2681 }
2682 
2683 void Assembler::movsd(Address dst, XMMRegister src) {
2684   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
2685   InstructionMark im(this);
2686   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
2687   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
2688   simd_prefix(src, xnoreg, dst, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
2689   emit_int8(0x11);
2690   emit_operand(src, dst);
2691 }
2692 
2693 void Assembler::movss(XMMRegister dst, XMMRegister src) {
2694   NOT_LP64(assert(VM_Version::supports_sse(), ""));
2695   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
2696   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
2697   emit_int8(0x10);
2698   emit_int8((unsigned char)(0xC0 | encode));
2699 }
2700 
2701 void Assembler::movss(XMMRegister dst, Address src) {
2702   NOT_LP64(assert(VM_Version::supports_sse(), ""));
2703   InstructionMark im(this);
2704   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
2705   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
2706   simd_prefix(dst, xnoreg, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
2707   emit_int8(0x10);
2708   emit_operand(dst, src);
2709 }
2710 
2711 void Assembler::movss(Address dst, XMMRegister src) {
2712   NOT_LP64(assert(VM_Version::supports_sse(), ""));
2713   InstructionMark im(this);
2714   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
2715   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
2716   simd_prefix(src, xnoreg, dst, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
2717   emit_int8(0x11);
2718   emit_operand(src, dst);
2719 }
2720 
2721 void Assembler::movswl(Register dst, Address src) { // movsxw
2722   InstructionMark im(this);
2723   prefix(src, dst);
2724   emit_int8(0x0F);
2725   emit_int8((unsigned char)0xBF);
2726   emit_operand(dst, src);
2727 }
2728 
2729 void Assembler::movswl(Register dst, Register src) { // movsxw
2730   int encode = prefix_and_encode(dst->encoding(), src->encoding());
2731   emit_int8(0x0F);
2732   emit_int8((unsigned char)0xBF);
2733   emit_int8((unsigned char)(0xC0 | encode));
2734 }
2735 
2736 void Assembler::movw(Address dst, int imm16) {
2737   InstructionMark im(this);
2738 
2739   emit_int8(0x66); // switch to 16-bit mode
2740   prefix(dst);
2741   emit_int8((unsigned char)0xC7);
2742   emit_operand(rax, dst, 2);
2743   emit_int16(imm16);
2744 }
2745 
2746 void Assembler::movw(Register dst, Address src) {
2747   InstructionMark im(this);
2748   emit_int8(0x66);
2749   prefix(src, dst);
2750   emit_int8((unsigned char)0x8B);
2751   emit_operand(dst, src);
2752 }
2753 
2754 void Assembler::movw(Address dst, Register src) {
2755   InstructionMark im(this);
2756   emit_int8(0x66);
2757   prefix(dst, src);
2758   emit_int8((unsigned char)0x89);
2759   emit_operand(src, dst);
2760 }
2761 
2762 void Assembler::movzbl(Register dst, Address src) { // movzxb
2763   InstructionMark im(this);
2764   prefix(src, dst);
2765   emit_int8(0x0F);
2766   emit_int8((unsigned char)0xB6);
2767   emit_operand(dst, src);
2768 }
2769 
2770 void Assembler::movzbl(Register dst, Register src) { // movzxb
2771   NOT_LP64(assert(src->has_byte_register(), "must have byte register"));
2772   int encode = prefix_and_encode(dst->encoding(), false, src->encoding(), true);
2773   emit_int8(0x0F);
2774   emit_int8((unsigned char)0xB6);
2775   emit_int8(0xC0 | encode);
2776 }
2777 
2778 void Assembler::movzwl(Register dst, Address src) { // movzxw
2779   InstructionMark im(this);
2780   prefix(src, dst);
2781   emit_int8(0x0F);
2782   emit_int8((unsigned char)0xB7);
2783   emit_operand(dst, src);
2784 }
2785 
2786 void Assembler::movzwl(Register dst, Register src) { // movzxw
2787   int encode = prefix_and_encode(dst->encoding(), src->encoding());
2788   emit_int8(0x0F);
2789   emit_int8((unsigned char)0xB7);
2790   emit_int8(0xC0 | encode);
2791 }
2792 
2793 void Assembler::mull(Address src) {
2794   InstructionMark im(this);
2795   prefix(src);
2796   emit_int8((unsigned char)0xF7);
2797   emit_operand(rsp, src);
2798 }
2799 
2800 void Assembler::mull(Register src) {
2801   int encode = prefix_and_encode(src->encoding());
2802   emit_int8((unsigned char)0xF7);
2803   emit_int8((unsigned char)(0xE0 | encode));
2804 }
2805 
2806 void Assembler::mulsd(XMMRegister dst, Address src) {
2807   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
2808   InstructionMark im(this);
2809   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
2810   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
2811   simd_prefix(dst, dst, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
2812   emit_int8(0x59);
2813   emit_operand(dst, src);
2814 }
2815 
2816 void Assembler::mulsd(XMMRegister dst, XMMRegister src) {
2817   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
2818   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
2819   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
2820   emit_int8(0x59);
2821   emit_int8((unsigned char)(0xC0 | encode));
2822 }
2823 
2824 void Assembler::mulss(XMMRegister dst, Address src) {
2825   NOT_LP64(assert(VM_Version::supports_sse(), ""));
2826   InstructionMark im(this);
2827   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
2828   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
2829   simd_prefix(dst, dst, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
2830   emit_int8(0x59);
2831   emit_operand(dst, src);
2832 }
2833 
2834 void Assembler::mulss(XMMRegister dst, XMMRegister src) {
2835   NOT_LP64(assert(VM_Version::supports_sse(), ""));
2836   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
2837   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
2838   emit_int8(0x59);
2839   emit_int8((unsigned char)(0xC0 | encode));
2840 }
2841 
2842 void Assembler::negl(Register dst) {
2843   int encode = prefix_and_encode(dst->encoding());
2844   emit_int8((unsigned char)0xF7);
2845   emit_int8((unsigned char)(0xD8 | encode));
2846 }
2847 
2848 void Assembler::nop(int i) {
2849 #ifdef ASSERT
2850   assert(i > 0, " ");
2851   // The fancy nops aren't currently recognized by debuggers making it a
2852   // pain to disassemble code while debugging. If asserts are on clearly
2853   // speed is not an issue so simply use the single byte traditional nop
2854   // to do alignment.
2855 
2856   for (; i > 0 ; i--) emit_int8((unsigned char)0x90);
2857   return;
2858 
2859 #endif // ASSERT
2860 
2861   if (UseAddressNop && VM_Version::is_intel()) {
2862     //
2863     // Using multi-bytes nops "0x0F 0x1F [address]" for Intel
2864     //  1: 0x90
2865     //  2: 0x66 0x90
2866     //  3: 0x66 0x66 0x90 (don't use "0x0F 0x1F 0x00" - need patching safe padding)
2867     //  4: 0x0F 0x1F 0x40 0x00
2868     //  5: 0x0F 0x1F 0x44 0x00 0x00
2869     //  6: 0x66 0x0F 0x1F 0x44 0x00 0x00
2870     //  7: 0x0F 0x1F 0x80 0x00 0x00 0x00 0x00
2871     //  8: 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00
2872     //  9: 0x66 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00
2873     // 10: 0x66 0x66 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00
2874     // 11: 0x66 0x66 0x66 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00
2875 
2876     // The rest coding is Intel specific - don't use consecutive address nops
2877 
2878     // 12: 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00 0x66 0x66 0x66 0x90
2879     // 13: 0x66 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00 0x66 0x66 0x66 0x90
2880     // 14: 0x66 0x66 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00 0x66 0x66 0x66 0x90
2881     // 15: 0x66 0x66 0x66 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00 0x66 0x66 0x66 0x90
2882 
2883     while(i >= 15) {
2884       // For Intel don't generate consecutive addess nops (mix with regular nops)
2885       i -= 15;
2886       emit_int8(0x66);   // size prefix
2887       emit_int8(0x66);   // size prefix
2888       emit_int8(0x66);   // size prefix
2889       addr_nop_8();
2890       emit_int8(0x66);   // size prefix
2891       emit_int8(0x66);   // size prefix
2892       emit_int8(0x66);   // size prefix
2893       emit_int8((unsigned char)0x90);
2894                          // nop
2895     }
2896     switch (i) {
2897       case 14:
2898         emit_int8(0x66); // size prefix
2899       case 13:
2900         emit_int8(0x66); // size prefix
2901       case 12:
2902         addr_nop_8();
2903         emit_int8(0x66); // size prefix
2904         emit_int8(0x66); // size prefix
2905         emit_int8(0x66); // size prefix
2906         emit_int8((unsigned char)0x90);
2907                          // nop
2908         break;
2909       case 11:
2910         emit_int8(0x66); // size prefix
2911       case 10:
2912         emit_int8(0x66); // size prefix
2913       case 9:
2914         emit_int8(0x66); // size prefix
2915       case 8:
2916         addr_nop_8();
2917         break;
2918       case 7:
2919         addr_nop_7();
2920         break;
2921       case 6:
2922         emit_int8(0x66); // size prefix
2923       case 5:
2924         addr_nop_5();
2925         break;
2926       case 4:
2927         addr_nop_4();
2928         break;
2929       case 3:
2930         // Don't use "0x0F 0x1F 0x00" - need patching safe padding
2931         emit_int8(0x66); // size prefix
2932       case 2:
2933         emit_int8(0x66); // size prefix
2934       case 1:
2935         emit_int8((unsigned char)0x90);
2936                          // nop
2937         break;
2938       default:
2939         assert(i == 0, " ");
2940     }
2941     return;
2942   }
2943   if (UseAddressNop && VM_Version::is_amd()) {
2944     //
2945     // Using multi-bytes nops "0x0F 0x1F [address]" for AMD.
2946     //  1: 0x90
2947     //  2: 0x66 0x90
2948     //  3: 0x66 0x66 0x90 (don't use "0x0F 0x1F 0x00" - need patching safe padding)
2949     //  4: 0x0F 0x1F 0x40 0x00
2950     //  5: 0x0F 0x1F 0x44 0x00 0x00
2951     //  6: 0x66 0x0F 0x1F 0x44 0x00 0x00
2952     //  7: 0x0F 0x1F 0x80 0x00 0x00 0x00 0x00
2953     //  8: 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00
2954     //  9: 0x66 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00
2955     // 10: 0x66 0x66 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00
2956     // 11: 0x66 0x66 0x66 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00
2957 
2958     // The rest coding is AMD specific - use consecutive address nops
2959 
2960     // 12: 0x66 0x0F 0x1F 0x44 0x00 0x00 0x66 0x0F 0x1F 0x44 0x00 0x00
2961     // 13: 0x0F 0x1F 0x80 0x00 0x00 0x00 0x00 0x66 0x0F 0x1F 0x44 0x00 0x00
2962     // 14: 0x0F 0x1F 0x80 0x00 0x00 0x00 0x00 0x0F 0x1F 0x80 0x00 0x00 0x00 0x00
2963     // 15: 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00 0x0F 0x1F 0x80 0x00 0x00 0x00 0x00
2964     // 16: 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00
2965     //     Size prefixes (0x66) are added for larger sizes
2966 
2967     while(i >= 22) {
2968       i -= 11;
2969       emit_int8(0x66); // size prefix
2970       emit_int8(0x66); // size prefix
2971       emit_int8(0x66); // size prefix
2972       addr_nop_8();
2973     }
2974     // Generate first nop for size between 21-12
2975     switch (i) {
2976       case 21:
2977         i -= 1;
2978         emit_int8(0x66); // size prefix
2979       case 20:
2980       case 19:
2981         i -= 1;
2982         emit_int8(0x66); // size prefix
2983       case 18:
2984       case 17:
2985         i -= 1;
2986         emit_int8(0x66); // size prefix
2987       case 16:
2988       case 15:
2989         i -= 8;
2990         addr_nop_8();
2991         break;
2992       case 14:
2993       case 13:
2994         i -= 7;
2995         addr_nop_7();
2996         break;
2997       case 12:
2998         i -= 6;
2999         emit_int8(0x66); // size prefix
3000         addr_nop_5();
3001         break;
3002       default:
3003         assert(i < 12, " ");
3004     }
3005 
3006     // Generate second nop for size between 11-1
3007     switch (i) {
3008       case 11:
3009         emit_int8(0x66); // size prefix
3010       case 10:
3011         emit_int8(0x66); // size prefix
3012       case 9:
3013         emit_int8(0x66); // size prefix
3014       case 8:
3015         addr_nop_8();
3016         break;
3017       case 7:
3018         addr_nop_7();
3019         break;
3020       case 6:
3021         emit_int8(0x66); // size prefix
3022       case 5:
3023         addr_nop_5();
3024         break;
3025       case 4:
3026         addr_nop_4();
3027         break;
3028       case 3:
3029         // Don't use "0x0F 0x1F 0x00" - need patching safe padding
3030         emit_int8(0x66); // size prefix
3031       case 2:
3032         emit_int8(0x66); // size prefix
3033       case 1:
3034         emit_int8((unsigned char)0x90);
3035                          // nop
3036         break;
3037       default:
3038         assert(i == 0, " ");
3039     }
3040     return;
3041   }
3042 
3043   // Using nops with size prefixes "0x66 0x90".
3044   // From AMD Optimization Guide:
3045   //  1: 0x90
3046   //  2: 0x66 0x90
3047   //  3: 0x66 0x66 0x90
3048   //  4: 0x66 0x66 0x66 0x90
3049   //  5: 0x66 0x66 0x90 0x66 0x90
3050   //  6: 0x66 0x66 0x90 0x66 0x66 0x90
3051   //  7: 0x66 0x66 0x66 0x90 0x66 0x66 0x90
3052   //  8: 0x66 0x66 0x66 0x90 0x66 0x66 0x66 0x90
3053   //  9: 0x66 0x66 0x90 0x66 0x66 0x90 0x66 0x66 0x90
3054   // 10: 0x66 0x66 0x66 0x90 0x66 0x66 0x90 0x66 0x66 0x90
3055   //
3056   while(i > 12) {
3057     i -= 4;
3058     emit_int8(0x66); // size prefix
3059     emit_int8(0x66);
3060     emit_int8(0x66);
3061     emit_int8((unsigned char)0x90);
3062                      // nop
3063   }
3064   // 1 - 12 nops
3065   if(i > 8) {
3066     if(i > 9) {
3067       i -= 1;
3068       emit_int8(0x66);
3069     }
3070     i -= 3;
3071     emit_int8(0x66);
3072     emit_int8(0x66);
3073     emit_int8((unsigned char)0x90);
3074   }
3075   // 1 - 8 nops
3076   if(i > 4) {
3077     if(i > 6) {
3078       i -= 1;
3079       emit_int8(0x66);
3080     }
3081     i -= 3;
3082     emit_int8(0x66);
3083     emit_int8(0x66);
3084     emit_int8((unsigned char)0x90);
3085   }
3086   switch (i) {
3087     case 4:
3088       emit_int8(0x66);
3089     case 3:
3090       emit_int8(0x66);
3091     case 2:
3092       emit_int8(0x66);
3093     case 1:
3094       emit_int8((unsigned char)0x90);
3095       break;
3096     default:
3097       assert(i == 0, " ");
3098   }
3099 }
3100 
3101 void Assembler::notl(Register dst) {
3102   int encode = prefix_and_encode(dst->encoding());
3103   emit_int8((unsigned char)0xF7);
3104   emit_int8((unsigned char)(0xD0 | encode));
3105 }
3106 
3107 void Assembler::orl(Address dst, int32_t imm32) {
3108   InstructionMark im(this);
3109   prefix(dst);
3110   emit_arith_operand(0x81, rcx, dst, imm32);
3111 }
3112 
3113 void Assembler::orl(Register dst, int32_t imm32) {
3114   prefix(dst);
3115   emit_arith(0x81, 0xC8, dst, imm32);
3116 }
3117 
3118 void Assembler::orl(Register dst, Address src) {
3119   InstructionMark im(this);
3120   prefix(src, dst);
3121   emit_int8(0x0B);
3122   emit_operand(dst, src);
3123 }
3124 
3125 void Assembler::orl(Register dst, Register src) {
3126   (void) prefix_and_encode(dst->encoding(), src->encoding());
3127   emit_arith(0x0B, 0xC0, dst, src);
3128 }
3129 
3130 void Assembler::orl(Address dst, Register src) {
3131   InstructionMark im(this);
3132   prefix(dst, src);
3133   emit_int8(0x09);
3134   emit_operand(src, dst);
3135 }
3136 
3137 void Assembler::packuswb(XMMRegister dst, Address src) {
3138   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
3139   assert((UseAVX > 0), "SSE mode requires address alignment 16 bytes");
3140   InstructionMark im(this);
3141   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
3142   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_32bit);
3143   simd_prefix(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3144   emit_int8(0x67);
3145   emit_operand(dst, src);
3146 }
3147 
3148 void Assembler::packuswb(XMMRegister dst, XMMRegister src) {
3149   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
3150   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
3151   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3152   emit_int8(0x67);
3153   emit_int8((unsigned char)(0xC0 | encode));
3154 }
3155 
3156 void Assembler::vpackuswb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
3157   assert(UseAVX > 0, "some form of AVX must be enabled");
3158   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
3159   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
3160   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3161   emit_int8(0x67);
3162   emit_int8((unsigned char)(0xC0 | encode));
3163 }
3164 
3165 void Assembler::vpermq(XMMRegister dst, XMMRegister src, int imm8, int vector_len) {
3166   assert(VM_Version::supports_avx2(), "");
3167   InstructionAttr attributes(vector_len, /* rex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
3168   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
3169   emit_int8(0x00);
3170   emit_int8(0xC0 | encode);
3171   emit_int8(imm8);
3172 }
3173 
3174 void Assembler::pause() {
3175   emit_int8((unsigned char)0xF3);
3176   emit_int8((unsigned char)0x90);
3177 }
3178 
3179 void Assembler::pcmpestri(XMMRegister dst, Address src, int imm8) {
3180   assert(VM_Version::supports_sse4_2(), "");
3181   InstructionMark im(this);
3182   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
3183   simd_prefix(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
3184   emit_int8(0x61);
3185   emit_operand(dst, src);
3186   emit_int8(imm8);
3187 }
3188 
3189 void Assembler::pcmpestri(XMMRegister dst, XMMRegister src, int imm8) {
3190   assert(VM_Version::supports_sse4_2(), "");
3191   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
3192   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
3193   emit_int8(0x61);
3194   emit_int8((unsigned char)(0xC0 | encode));
3195   emit_int8(imm8);
3196 }
3197 
3198 // In this context, the dst vector contains the components that are equal, non equal components are zeroed in dst
3199 void Assembler::pcmpeqb(XMMRegister dst, XMMRegister src) {
3200   assert(VM_Version::supports_sse2(), "");
3201   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
3202   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3203   emit_int8(0x74);
3204   emit_int8((unsigned char)(0xC0 | encode));
3205 }
3206 
3207 // In this context, the dst vector contains the components that are equal, non equal components are zeroed in dst
3208 void Assembler::vpcmpeqb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
3209   assert(VM_Version::supports_avx(), "");
3210   InstructionAttr attributes(vector_len, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
3211   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
3212   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3213   emit_int8(0x74);
3214   emit_int8((unsigned char)(0xC0 | encode));
3215 }
3216 
3217 // In this context, kdst is written the mask used to process the equal components
3218 void Assembler::evpcmpeqb(KRegister kdst, XMMRegister nds, XMMRegister src, int vector_len) {
3219   assert(VM_Version::supports_avx512bw(), "");
3220   InstructionAttr attributes(vector_len, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ true);
3221   attributes.set_is_evex_instruction();
3222   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
3223   int encode = vex_prefix_and_encode(kdst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3224   emit_int8(0x74);
3225   emit_int8((unsigned char)(0xC0 | encode));
3226 }
3227 
3228 void Assembler::evpcmpeqb(KRegister kdst, XMMRegister nds, Address src, int vector_len) {
3229   assert(VM_Version::supports_avx512bw(), "");
3230   InstructionMark im(this);
3231   InstructionAttr attributes(vector_len, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ true);
3232   attributes.set_is_evex_instruction();
3233   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
3234   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
3235   int dst_enc = kdst->encoding();
3236   vex_prefix(src, nds_enc, dst_enc, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3237   emit_int8(0x74);
3238   emit_operand(as_Register(dst_enc), src);
3239 }
3240 
3241 // In this context, the dst vector contains the components that are equal, non equal components are zeroed in dst
3242 void Assembler::pcmpeqw(XMMRegister dst, XMMRegister src) {
3243   assert(VM_Version::supports_sse2(), "");
3244   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
3245   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3246   emit_int8(0x75);
3247   emit_int8((unsigned char)(0xC0 | encode));
3248 }
3249 
3250 // In this context, the dst vector contains the components that are equal, non equal components are zeroed in dst
3251 void Assembler::vpcmpeqw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
3252   assert(VM_Version::supports_avx(), "");
3253   InstructionAttr attributes(vector_len, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
3254   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
3255   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3256   emit_int8(0x75);
3257   emit_int8((unsigned char)(0xC0 | encode));
3258 }
3259 
3260 // In this context, kdst is written the mask used to process the equal components
3261 void Assembler::evpcmpeqw(KRegister kdst, XMMRegister nds, XMMRegister src, int vector_len) {
3262   assert(VM_Version::supports_avx512bw(), "");
3263   InstructionAttr attributes(vector_len, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ true);
3264   attributes.set_is_evex_instruction();
3265   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
3266   int encode = vex_prefix_and_encode(kdst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3267   emit_int8(0x75);
3268   emit_int8((unsigned char)(0xC0 | encode));
3269 }
3270 
3271 void Assembler::evpcmpeqw(KRegister kdst, XMMRegister nds, Address src, int vector_len) {
3272   assert(VM_Version::supports_avx512bw(), "");
3273   InstructionMark im(this);
3274   InstructionAttr attributes(vector_len, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ true);
3275   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
3276   attributes.set_is_evex_instruction();
3277   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
3278   int dst_enc = kdst->encoding();
3279   vex_prefix(src, nds_enc, dst_enc, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3280   emit_int8(0x75);
3281   emit_operand(as_Register(dst_enc), src);
3282 }
3283 
3284 // In this context, the dst vector contains the components that are equal, non equal components are zeroed in dst
3285 void Assembler::pcmpeqd(XMMRegister dst, XMMRegister src) {
3286   assert(VM_Version::supports_sse2(), "");
3287   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
3288   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3289   emit_int8(0x76);
3290   emit_int8((unsigned char)(0xC0 | encode));
3291 }
3292 
3293 // In this context, the dst vector contains the components that are equal, non equal components are zeroed in dst
3294 void Assembler::vpcmpeqd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
3295   assert(VM_Version::supports_avx(), "");
3296   InstructionAttr attributes(vector_len, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
3297   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
3298   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3299   emit_int8(0x76);
3300   emit_int8((unsigned char)(0xC0 | encode));
3301 }
3302 
3303 // In this context, kdst is written the mask used to process the equal components
3304 void Assembler::evpcmpeqd(KRegister kdst, XMMRegister nds, XMMRegister src, int vector_len) {
3305   assert(VM_Version::supports_evex(), "");
3306   InstructionAttr attributes(vector_len, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
3307   attributes.set_is_evex_instruction();
3308   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
3309   int encode = vex_prefix_and_encode(kdst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3310   emit_int8(0x76);
3311   emit_int8((unsigned char)(0xC0 | encode));
3312 }
3313 
3314 void Assembler::evpcmpeqd(KRegister kdst, XMMRegister nds, Address src, int vector_len) {
3315   assert(VM_Version::supports_evex(), "");
3316   InstructionMark im(this);
3317   InstructionAttr attributes(vector_len, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
3318   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_32bit);
3319   attributes.set_is_evex_instruction();
3320   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
3321   int dst_enc = kdst->encoding();
3322   vex_prefix(src, nds_enc, dst_enc, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3323   emit_int8(0x76);
3324   emit_operand(as_Register(dst_enc), src);
3325 }
3326 
3327 // In this context, the dst vector contains the components that are equal, non equal components are zeroed in dst
3328 void Assembler::pcmpeqq(XMMRegister dst, XMMRegister src) {
3329   assert(VM_Version::supports_sse4_1(), "");
3330   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
3331   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
3332   emit_int8(0x29);
3333   emit_int8((unsigned char)(0xC0 | encode));
3334 }
3335 
3336 // In this context, the dst vector contains the components that are equal, non equal components are zeroed in dst
3337 void Assembler::vpcmpeqq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
3338   assert(VM_Version::supports_avx(), "");
3339   InstructionAttr attributes(vector_len, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
3340   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
3341   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
3342   emit_int8(0x29);
3343   emit_int8((unsigned char)(0xC0 | encode));
3344 }
3345 
3346 // In this context, kdst is written the mask used to process the equal components
3347 void Assembler::evpcmpeqq(KRegister kdst, XMMRegister nds, XMMRegister src, int vector_len) {
3348   assert(VM_Version::supports_evex(), "");
3349   InstructionAttr attributes(vector_len, /* rex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
3350   attributes.set_is_evex_instruction();
3351   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
3352   int encode = vex_prefix_and_encode(kdst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
3353   emit_int8(0x29);
3354   emit_int8((unsigned char)(0xC0 | encode));
3355 }
3356 
3357 // In this context, kdst is written the mask used to process the equal components
3358 void Assembler::evpcmpeqq(KRegister kdst, XMMRegister nds, Address src, int vector_len) {
3359   assert(VM_Version::supports_evex(), "");
3360   InstructionMark im(this);
3361   InstructionAttr attributes(vector_len, /* rex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
3362   attributes.set_is_evex_instruction();
3363   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_64bit);
3364   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
3365   int dst_enc = kdst->encoding();
3366   vex_prefix(src, nds_enc, dst_enc, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
3367   emit_int8(0x29);
3368   emit_operand(as_Register(dst_enc), src);
3369 }
3370 
3371 void Assembler::pmovmskb(Register dst, XMMRegister src) {
3372   assert(VM_Version::supports_sse2(), "");
3373   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
3374   int encode = simd_prefix_and_encode(as_XMMRegister(dst->encoding()), xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3375   emit_int8((unsigned char)0xD7);
3376   emit_int8((unsigned char)(0xC0 | encode));
3377 }
3378 
3379 void Assembler::vpmovmskb(Register dst, XMMRegister src) {
3380   assert(VM_Version::supports_avx2(), "");
3381   InstructionAttr attributes(AVX_256bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
3382   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3383   emit_int8((unsigned char)0xD7);
3384   emit_int8((unsigned char)(0xC0 | encode));
3385 }
3386 
3387 void Assembler::pextrd(Register dst, XMMRegister src, int imm8) {
3388   assert(VM_Version::supports_sse4_1(), "");
3389   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ true, /* uses_vl */ false);
3390   int encode = simd_prefix_and_encode(src, xnoreg, as_XMMRegister(dst->encoding()), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
3391   emit_int8(0x16);
3392   emit_int8((unsigned char)(0xC0 | encode));
3393   emit_int8(imm8);
3394 }
3395 
3396 void Assembler::pextrd(Address dst, XMMRegister src, int imm8) {
3397   assert(VM_Version::supports_sse4_1(), "");
3398   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ true, /* uses_vl */ false);
3399   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
3400   simd_prefix(src, xnoreg, dst, VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
3401   emit_int8(0x16);
3402   emit_operand(src, dst);
3403   emit_int8(imm8);
3404 }
3405 
3406 void Assembler::pextrq(Register dst, XMMRegister src, int imm8) {
3407   assert(VM_Version::supports_sse4_1(), "");
3408   InstructionAttr attributes(AVX_128bit, /* rex_w */ true, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ true, /* uses_vl */ false);
3409   int encode = simd_prefix_and_encode(src, xnoreg, as_XMMRegister(dst->encoding()), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
3410   emit_int8(0x16);
3411   emit_int8((unsigned char)(0xC0 | encode));
3412   emit_int8(imm8);
3413 }
3414 
3415 void Assembler::pextrq(Address dst, XMMRegister src, int imm8) {
3416   assert(VM_Version::supports_sse4_1(), "");
3417   InstructionAttr attributes(AVX_128bit, /* rex_w */ true, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ true, /* uses_vl */ false);
3418   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
3419   simd_prefix(src, xnoreg, dst, VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
3420   emit_int8(0x16);
3421   emit_operand(src, dst);
3422   emit_int8(imm8);
3423 }
3424 
3425 void Assembler::pextrw(Register dst, XMMRegister src, int imm8) {
3426   assert(VM_Version::supports_sse2(), "");
3427   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ false);
3428   int encode = simd_prefix_and_encode(as_XMMRegister(dst->encoding()), xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3429   emit_int8((unsigned char)0xC5);
3430   emit_int8((unsigned char)(0xC0 | encode));
3431   emit_int8(imm8);
3432 }
3433 
3434 void Assembler::pextrw(Address dst, XMMRegister src, int imm8) {
3435   assert(VM_Version::supports_sse4_1(), "");
3436   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ false);
3437   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_16bit);
3438   simd_prefix(src, xnoreg, dst, VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
3439   emit_int8((unsigned char)0x15);
3440   emit_operand(src, dst);
3441   emit_int8(imm8);
3442 }
3443 
3444 void Assembler::pextrb(Address dst, XMMRegister src, int imm8) {
3445   assert(VM_Version::supports_sse4_1(), "");
3446   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ false);
3447   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_8bit);
3448   simd_prefix(src, xnoreg, dst, VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
3449   emit_int8(0x14);
3450   emit_operand(src, dst);
3451   emit_int8(imm8);
3452 }
3453 
3454 void Assembler::pinsrd(XMMRegister dst, Register src, int imm8) {
3455   assert(VM_Version::supports_sse4_1(), "");
3456   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ true, /* uses_vl */ false);
3457   int encode = simd_prefix_and_encode(dst, dst, as_XMMRegister(src->encoding()), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
3458   emit_int8(0x22);
3459   emit_int8((unsigned char)(0xC0 | encode));
3460   emit_int8(imm8);
3461 }
3462 
3463 void Assembler::pinsrd(XMMRegister dst, Address src, int imm8) {
3464   assert(VM_Version::supports_sse4_1(), "");
3465   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ true, /* uses_vl */ false);
3466   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
3467   simd_prefix(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
3468   emit_int8(0x22);
3469   emit_operand(dst,src);
3470   emit_int8(imm8);
3471 }
3472 
3473 void Assembler::pinsrq(XMMRegister dst, Register src, int imm8) {
3474   assert(VM_Version::supports_sse4_1(), "");
3475   InstructionAttr attributes(AVX_128bit, /* rex_w */ true, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ true, /* uses_vl */ false);
3476   int encode = simd_prefix_and_encode(dst, dst, as_XMMRegister(src->encoding()), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
3477   emit_int8(0x22);
3478   emit_int8((unsigned char)(0xC0 | encode));
3479   emit_int8(imm8);
3480 }
3481 
3482 void Assembler::pinsrq(XMMRegister dst, Address src, int imm8) {
3483   assert(VM_Version::supports_sse4_1(), "");
3484   InstructionAttr attributes(AVX_128bit, /* rex_w */ true, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ true, /* uses_vl */ false);
3485   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
3486   simd_prefix(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
3487   emit_int8(0x22);
3488   emit_operand(dst, src);
3489   emit_int8(imm8);
3490 }
3491 
3492 void Assembler::pinsrw(XMMRegister dst, Register src, int imm8) {
3493   assert(VM_Version::supports_sse2(), "");
3494   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ false);
3495   int encode = simd_prefix_and_encode(dst, dst, as_XMMRegister(src->encoding()), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3496   emit_int8((unsigned char)0xC4);
3497   emit_int8((unsigned char)(0xC0 | encode));
3498   emit_int8(imm8);
3499 }
3500 
3501 void Assembler::pinsrw(XMMRegister dst, Address src, int imm8) {
3502   assert(VM_Version::supports_sse2(), "");
3503   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ false);
3504   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_16bit);
3505   simd_prefix(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3506   emit_int8((unsigned char)0xC4);
3507   emit_operand(dst, src);
3508   emit_int8(imm8);
3509 }
3510 
3511 void Assembler::pinsrb(XMMRegister dst, Address src, int imm8) {
3512   assert(VM_Version::supports_sse4_1(), "");
3513   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ false);
3514   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_8bit);
3515   simd_prefix(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
3516   emit_int8(0x20);
3517   emit_operand(dst, src);
3518   emit_int8(imm8);
3519 }
3520 
3521 void Assembler::pmovzxbw(XMMRegister dst, Address src) {
3522   assert(VM_Version::supports_sse4_1(), "");
3523   InstructionMark im(this);
3524   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ false);
3525   attributes.set_address_attributes(/* tuple_type */ EVEX_HVM, /* input_size_in_bits */ EVEX_NObit);
3526   simd_prefix(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
3527   emit_int8(0x30);
3528   emit_operand(dst, src);
3529 }
3530 
3531 void Assembler::pmovzxbw(XMMRegister dst, XMMRegister src) {
3532   assert(VM_Version::supports_sse4_1(), "");
3533   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ false);
3534   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
3535   emit_int8(0x30);
3536   emit_int8((unsigned char)(0xC0 | encode));
3537 }
3538 
3539 void Assembler::vpmovzxbw(XMMRegister dst, Address src, int vector_len) {
3540   assert(VM_Version::supports_avx(), "");
3541   InstructionMark im(this);
3542   assert(dst != xnoreg, "sanity");
3543   InstructionAttr attributes(vector_len, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ false);
3544   attributes.set_address_attributes(/* tuple_type */ EVEX_HVM, /* input_size_in_bits */ EVEX_NObit);
3545   vex_prefix(src, 0, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
3546   emit_int8(0x30);
3547   emit_operand(dst, src);
3548 }
3549 
3550 // generic
3551 void Assembler::pop(Register dst) {
3552   int encode = prefix_and_encode(dst->encoding());
3553   emit_int8(0x58 | encode);
3554 }
3555 
3556 void Assembler::popcntl(Register dst, Address src) {
3557   assert(VM_Version::supports_popcnt(), "must support");
3558   InstructionMark im(this);
3559   emit_int8((unsigned char)0xF3);
3560   prefix(src, dst);
3561   emit_int8(0x0F);
3562   emit_int8((unsigned char)0xB8);
3563   emit_operand(dst, src);
3564 }
3565 
3566 void Assembler::popcntl(Register dst, Register src) {
3567   assert(VM_Version::supports_popcnt(), "must support");
3568   emit_int8((unsigned char)0xF3);
3569   int encode = prefix_and_encode(dst->encoding(), src->encoding());
3570   emit_int8(0x0F);
3571   emit_int8((unsigned char)0xB8);
3572   emit_int8((unsigned char)(0xC0 | encode));
3573 }
3574 
3575 void Assembler::popf() {
3576   emit_int8((unsigned char)0x9D);
3577 }
3578 
3579 #ifndef _LP64 // no 32bit push/pop on amd64
3580 void Assembler::popl(Address dst) {
3581   // NOTE: this will adjust stack by 8byte on 64bits
3582   InstructionMark im(this);
3583   prefix(dst);
3584   emit_int8((unsigned char)0x8F);
3585   emit_operand(rax, dst);
3586 }
3587 #endif
3588 
3589 void Assembler::prefetch_prefix(Address src) {
3590   prefix(src);
3591   emit_int8(0x0F);
3592 }
3593 
3594 void Assembler::prefetchnta(Address src) {
3595   NOT_LP64(assert(VM_Version::supports_sse(), "must support"));
3596   InstructionMark im(this);
3597   prefetch_prefix(src);
3598   emit_int8(0x18);
3599   emit_operand(rax, src); // 0, src
3600 }
3601 
3602 void Assembler::prefetchr(Address src) {
3603   assert(VM_Version::supports_3dnow_prefetch(), "must support");
3604   InstructionMark im(this);
3605   prefetch_prefix(src);
3606   emit_int8(0x0D);
3607   emit_operand(rax, src); // 0, src
3608 }
3609 
3610 void Assembler::prefetcht0(Address src) {
3611   NOT_LP64(assert(VM_Version::supports_sse(), "must support"));
3612   InstructionMark im(this);
3613   prefetch_prefix(src);
3614   emit_int8(0x18);
3615   emit_operand(rcx, src); // 1, src
3616 }
3617 
3618 void Assembler::prefetcht1(Address src) {
3619   NOT_LP64(assert(VM_Version::supports_sse(), "must support"));
3620   InstructionMark im(this);
3621   prefetch_prefix(src);
3622   emit_int8(0x18);
3623   emit_operand(rdx, src); // 2, src
3624 }
3625 
3626 void Assembler::prefetcht2(Address src) {
3627   NOT_LP64(assert(VM_Version::supports_sse(), "must support"));
3628   InstructionMark im(this);
3629   prefetch_prefix(src);
3630   emit_int8(0x18);
3631   emit_operand(rbx, src); // 3, src
3632 }
3633 
3634 void Assembler::prefetchw(Address src) {
3635   assert(VM_Version::supports_3dnow_prefetch(), "must support");
3636   InstructionMark im(this);
3637   prefetch_prefix(src);
3638   emit_int8(0x0D);
3639   emit_operand(rcx, src); // 1, src
3640 }
3641 
3642 void Assembler::prefix(Prefix p) {
3643   emit_int8(p);
3644 }
3645 
3646 void Assembler::pshufb(XMMRegister dst, XMMRegister src) {
3647   assert(VM_Version::supports_ssse3(), "");
3648   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ false);
3649   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
3650   emit_int8(0x00);
3651   emit_int8((unsigned char)(0xC0 | encode));
3652 }
3653 
3654 void Assembler::pshufb(XMMRegister dst, Address src) {
3655   assert(VM_Version::supports_ssse3(), "");
3656   InstructionMark im(this);
3657   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ false);
3658   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
3659   simd_prefix(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
3660   emit_int8(0x00);
3661   emit_operand(dst, src);
3662 }
3663 
3664 void Assembler::pshufd(XMMRegister dst, XMMRegister src, int mode) {
3665   assert(isByte(mode), "invalid value");
3666   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
3667   int vector_len = VM_Version::supports_avx512novl() ? AVX_512bit : AVX_128bit;
3668   InstructionAttr attributes(vector_len, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
3669   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3670   emit_int8(0x70);
3671   emit_int8((unsigned char)(0xC0 | encode));
3672   emit_int8(mode & 0xFF);
3673 }
3674 
3675 void Assembler::pshufd(XMMRegister dst, Address src, int mode) {
3676   assert(isByte(mode), "invalid value");
3677   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
3678   assert((UseAVX > 0), "SSE mode requires address alignment 16 bytes");
3679   InstructionMark im(this);
3680   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
3681   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_32bit);
3682   simd_prefix(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3683   emit_int8(0x70);
3684   emit_operand(dst, src);
3685   emit_int8(mode & 0xFF);
3686 }
3687 
3688 void Assembler::pshuflw(XMMRegister dst, XMMRegister src, int mode) {
3689   assert(isByte(mode), "invalid value");
3690   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
3691   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ false);
3692   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
3693   emit_int8(0x70);
3694   emit_int8((unsigned char)(0xC0 | encode));
3695   emit_int8(mode & 0xFF);
3696 }
3697 
3698 void Assembler::pshuflw(XMMRegister dst, Address src, int mode) {
3699   assert(isByte(mode), "invalid value");
3700   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
3701   assert((UseAVX > 0), "SSE mode requires address alignment 16 bytes");
3702   InstructionMark im(this);
3703   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ false);
3704   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
3705   simd_prefix(dst, xnoreg, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
3706   emit_int8(0x70);
3707   emit_operand(dst, src);
3708   emit_int8(mode & 0xFF);
3709 }
3710 
3711 void Assembler::psrldq(XMMRegister dst, int shift) {
3712   // Shift left 128 bit value in dst XMMRegister by shift number of bytes.
3713   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
3714   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ false);
3715   // XMM3 is for /3 encoding: 66 0F 73 /3 ib
3716   int encode = simd_prefix_and_encode(xmm3, dst, dst, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3717   emit_int8(0x73);
3718   emit_int8((unsigned char)(0xC0 | encode));
3719   emit_int8(shift);
3720 }
3721 
3722 void Assembler::pslldq(XMMRegister dst, int shift) {
3723   // Shift left 128 bit value in dst XMMRegister by shift number of bytes.
3724   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
3725   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ false);
3726   // XMM7 is for /7 encoding: 66 0F 73 /7 ib
3727   int encode = simd_prefix_and_encode(xmm7, dst, dst, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3728   emit_int8(0x73);
3729   emit_int8((unsigned char)(0xC0 | encode));
3730   emit_int8(shift);
3731 }
3732 
3733 void Assembler::ptest(XMMRegister dst, Address src) {
3734   assert(VM_Version::supports_sse4_1(), "");
3735   assert((UseAVX > 0), "SSE mode requires address alignment 16 bytes");
3736   InstructionMark im(this);
3737   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
3738   simd_prefix(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
3739   emit_int8(0x17);
3740   emit_operand(dst, src);
3741 }
3742 
3743 void Assembler::ptest(XMMRegister dst, XMMRegister src) {
3744   assert(VM_Version::supports_sse4_1(), "");
3745   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
3746   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
3747   emit_int8(0x17);
3748   emit_int8((unsigned char)(0xC0 | encode));
3749 }
3750 
3751 void Assembler::vptest(XMMRegister dst, Address src) {
3752   assert(VM_Version::supports_avx(), "");
3753   InstructionMark im(this);
3754   InstructionAttr attributes(AVX_256bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
3755   assert(dst != xnoreg, "sanity");
3756   // swap src<->dst for encoding
3757   vex_prefix(src, 0, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
3758   emit_int8(0x17);
3759   emit_operand(dst, src);
3760 }
3761 
3762 void Assembler::vptest(XMMRegister dst, XMMRegister src) {
3763   assert(VM_Version::supports_avx(), "");
3764   InstructionAttr attributes(AVX_256bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
3765   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
3766   emit_int8(0x17);
3767   emit_int8((unsigned char)(0xC0 | encode));
3768 }
3769 
3770 void Assembler::punpcklbw(XMMRegister dst, Address src) {
3771   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
3772   assert((UseAVX > 0), "SSE mode requires address alignment 16 bytes");
3773   InstructionMark im(this);
3774   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_vlbw, /* no_mask_reg */ false, /* uses_vl */ true);
3775   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
3776   simd_prefix(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3777   emit_int8(0x60);
3778   emit_operand(dst, src);
3779 }
3780 
3781 void Assembler::punpcklbw(XMMRegister dst, XMMRegister src) {
3782   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
3783   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_vlbw, /* no_mask_reg */ false, /* uses_vl */ true);
3784   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3785   emit_int8(0x60);
3786   emit_int8((unsigned char)(0xC0 | encode));
3787 }
3788 
3789 void Assembler::punpckldq(XMMRegister dst, Address src) {
3790   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
3791   assert((UseAVX > 0), "SSE mode requires address alignment 16 bytes");
3792   InstructionMark im(this);
3793   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
3794   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_32bit);
3795   simd_prefix(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3796   emit_int8(0x62);
3797   emit_operand(dst, src);
3798 }
3799 
3800 void Assembler::punpckldq(XMMRegister dst, XMMRegister src) {
3801   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
3802   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
3803   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3804   emit_int8(0x62);
3805   emit_int8((unsigned char)(0xC0 | encode));
3806 }
3807 
3808 void Assembler::punpcklqdq(XMMRegister dst, XMMRegister src) {
3809   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
3810   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
3811   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
3812   emit_int8(0x6C);
3813   emit_int8((unsigned char)(0xC0 | encode));
3814 }
3815 
3816 void Assembler::push(int32_t imm32) {
3817   // in 64bits we push 64bits onto the stack but only
3818   // take a 32bit immediate
3819   emit_int8(0x68);
3820   emit_int32(imm32);
3821 }
3822 
3823 void Assembler::push(Register src) {
3824   int encode = prefix_and_encode(src->encoding());
3825 
3826   emit_int8(0x50 | encode);
3827 }
3828 
3829 void Assembler::pushf() {
3830   emit_int8((unsigned char)0x9C);
3831 }
3832 
3833 #ifndef _LP64 // no 32bit push/pop on amd64
3834 void Assembler::pushl(Address src) {
3835   // Note this will push 64bit on 64bit
3836   InstructionMark im(this);
3837   prefix(src);
3838   emit_int8((unsigned char)0xFF);
3839   emit_operand(rsi, src);
3840 }
3841 #endif
3842 
3843 void Assembler::rcll(Register dst, int imm8) {
3844   assert(isShiftCount(imm8), "illegal shift count");
3845   int encode = prefix_and_encode(dst->encoding());
3846   if (imm8 == 1) {
3847     emit_int8((unsigned char)0xD1);
3848     emit_int8((unsigned char)(0xD0 | encode));
3849   } else {
3850     emit_int8((unsigned char)0xC1);
3851     emit_int8((unsigned char)0xD0 | encode);
3852     emit_int8(imm8);
3853   }
3854 }
3855 
3856 void Assembler::rcpps(XMMRegister dst, XMMRegister src) {
3857   NOT_LP64(assert(VM_Version::supports_sse(), ""));
3858   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
3859   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
3860   emit_int8(0x53);
3861   emit_int8((unsigned char)(0xC0 | encode));
3862 }
3863 
3864 void Assembler::rcpss(XMMRegister dst, XMMRegister src) {
3865   NOT_LP64(assert(VM_Version::supports_sse(), ""));
3866   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
3867   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
3868   emit_int8(0x53);
3869   emit_int8((unsigned char)(0xC0 | encode));
3870 }
3871 
3872 void Assembler::rdtsc() {
3873   emit_int8((unsigned char)0x0F);
3874   emit_int8((unsigned char)0x31);
3875 }
3876 
3877 // copies data from [esi] to [edi] using rcx pointer sized words
3878 // generic
3879 void Assembler::rep_mov() {
3880   emit_int8((unsigned char)0xF3);
3881   // MOVSQ
3882   LP64_ONLY(prefix(REX_W));
3883   emit_int8((unsigned char)0xA5);
3884 }
3885 
3886 // sets rcx bytes with rax, value at [edi]
3887 void Assembler::rep_stosb() {
3888   emit_int8((unsigned char)0xF3); // REP
3889   LP64_ONLY(prefix(REX_W));
3890   emit_int8((unsigned char)0xAA); // STOSB
3891 }
3892 
3893 // sets rcx pointer sized words with rax, value at [edi]
3894 // generic
3895 void Assembler::rep_stos() {
3896   emit_int8((unsigned char)0xF3); // REP
3897   LP64_ONLY(prefix(REX_W));       // LP64:STOSQ, LP32:STOSD
3898   emit_int8((unsigned char)0xAB);
3899 }
3900 
3901 // scans rcx pointer sized words at [edi] for occurance of rax,
3902 // generic
3903 void Assembler::repne_scan() { // repne_scan
3904   emit_int8((unsigned char)0xF2);
3905   // SCASQ
3906   LP64_ONLY(prefix(REX_W));
3907   emit_int8((unsigned char)0xAF);
3908 }
3909 
3910 #ifdef _LP64
3911 // scans rcx 4 byte words at [edi] for occurance of rax,
3912 // generic
3913 void Assembler::repne_scanl() { // repne_scan
3914   emit_int8((unsigned char)0xF2);
3915   // SCASL
3916   emit_int8((unsigned char)0xAF);
3917 }
3918 #endif
3919 
3920 void Assembler::ret(int imm16) {
3921   if (imm16 == 0) {
3922     emit_int8((unsigned char)0xC3);
3923   } else {
3924     emit_int8((unsigned char)0xC2);
3925     emit_int16(imm16);
3926   }
3927 }
3928 
3929 void Assembler::sahf() {
3930 #ifdef _LP64
3931   // Not supported in 64bit mode
3932   ShouldNotReachHere();
3933 #endif
3934   emit_int8((unsigned char)0x9E);
3935 }
3936 
3937 void Assembler::sarl(Register dst, int imm8) {
3938   int encode = prefix_and_encode(dst->encoding());
3939   assert(isShiftCount(imm8), "illegal shift count");
3940   if (imm8 == 1) {
3941     emit_int8((unsigned char)0xD1);
3942     emit_int8((unsigned char)(0xF8 | encode));
3943   } else {
3944     emit_int8((unsigned char)0xC1);
3945     emit_int8((unsigned char)(0xF8 | encode));
3946     emit_int8(imm8);
3947   }
3948 }
3949 
3950 void Assembler::sarl(Register dst) {
3951   int encode = prefix_and_encode(dst->encoding());
3952   emit_int8((unsigned char)0xD3);
3953   emit_int8((unsigned char)(0xF8 | encode));
3954 }
3955 
3956 void Assembler::sbbl(Address dst, int32_t imm32) {
3957   InstructionMark im(this);
3958   prefix(dst);
3959   emit_arith_operand(0x81, rbx, dst, imm32);
3960 }
3961 
3962 void Assembler::sbbl(Register dst, int32_t imm32) {
3963   prefix(dst);
3964   emit_arith(0x81, 0xD8, dst, imm32);
3965 }
3966 
3967 
3968 void Assembler::sbbl(Register dst, Address src) {
3969   InstructionMark im(this);
3970   prefix(src, dst);
3971   emit_int8(0x1B);
3972   emit_operand(dst, src);
3973 }
3974 
3975 void Assembler::sbbl(Register dst, Register src) {
3976   (void) prefix_and_encode(dst->encoding(), src->encoding());
3977   emit_arith(0x1B, 0xC0, dst, src);
3978 }
3979 
3980 void Assembler::setb(Condition cc, Register dst) {
3981   assert(0 <= cc && cc < 16, "illegal cc");
3982   int encode = prefix_and_encode(dst->encoding(), true);
3983   emit_int8(0x0F);
3984   emit_int8((unsigned char)0x90 | cc);
3985   emit_int8((unsigned char)(0xC0 | encode));
3986 }
3987 
3988 void Assembler::palignr(XMMRegister dst, XMMRegister src, int imm8) {
3989   assert(VM_Version::supports_ssse3(), "");
3990   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ false, /* uses_vl */ false);
3991   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
3992   emit_int8((unsigned char)0x0F);
3993   emit_int8((unsigned char)(0xC0 | encode));
3994   emit_int8(imm8);
3995 }
3996 
3997 void Assembler::pblendw(XMMRegister dst, XMMRegister src, int imm8) {
3998   assert(VM_Version::supports_sse4_1(), "");
3999   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
4000   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
4001   emit_int8((unsigned char)0x0E);
4002   emit_int8((unsigned char)(0xC0 | encode));
4003   emit_int8(imm8);
4004 }
4005 
4006 void Assembler::sha1rnds4(XMMRegister dst, XMMRegister src, int imm8) {
4007   assert(VM_Version::supports_sha(), "");
4008   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
4009   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_NONE, VEX_OPCODE_0F_3A, &attributes);
4010   emit_int8((unsigned char)0xCC);
4011   emit_int8((unsigned char)(0xC0 | encode));
4012   emit_int8((unsigned char)imm8);
4013 }
4014 
4015 void Assembler::sha1nexte(XMMRegister dst, XMMRegister src) {
4016   assert(VM_Version::supports_sha(), "");
4017   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
4018   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
4019   emit_int8((unsigned char)0xC8);
4020   emit_int8((unsigned char)(0xC0 | encode));
4021 }
4022 
4023 void Assembler::sha1msg1(XMMRegister dst, XMMRegister src) {
4024   assert(VM_Version::supports_sha(), "");
4025   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
4026   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
4027   emit_int8((unsigned char)0xC9);
4028   emit_int8((unsigned char)(0xC0 | encode));
4029 }
4030 
4031 void Assembler::sha1msg2(XMMRegister dst, XMMRegister src) {
4032   assert(VM_Version::supports_sha(), "");
4033   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
4034   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
4035   emit_int8((unsigned char)0xCA);
4036   emit_int8((unsigned char)(0xC0 | encode));
4037 }
4038 
4039 // xmm0 is implicit additional source to this instruction.
4040 void Assembler::sha256rnds2(XMMRegister dst, XMMRegister src) {
4041   assert(VM_Version::supports_sha(), "");
4042   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
4043   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
4044   emit_int8((unsigned char)0xCB);
4045   emit_int8((unsigned char)(0xC0 | encode));
4046 }
4047 
4048 void Assembler::sha256msg1(XMMRegister dst, XMMRegister src) {
4049   assert(VM_Version::supports_sha(), "");
4050   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
4051   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
4052   emit_int8((unsigned char)0xCC);
4053   emit_int8((unsigned char)(0xC0 | encode));
4054 }
4055 
4056 void Assembler::sha256msg2(XMMRegister dst, XMMRegister src) {
4057   assert(VM_Version::supports_sha(), "");
4058   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
4059   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
4060   emit_int8((unsigned char)0xCD);
4061   emit_int8((unsigned char)(0xC0 | encode));
4062 }
4063 
4064 
4065 void Assembler::shll(Register dst, int imm8) {
4066   assert(isShiftCount(imm8), "illegal shift count");
4067   int encode = prefix_and_encode(dst->encoding());
4068   if (imm8 == 1 ) {
4069     emit_int8((unsigned char)0xD1);
4070     emit_int8((unsigned char)(0xE0 | encode));
4071   } else {
4072     emit_int8((unsigned char)0xC1);
4073     emit_int8((unsigned char)(0xE0 | encode));
4074     emit_int8(imm8);
4075   }
4076 }
4077 
4078 void Assembler::shll(Register dst) {
4079   int encode = prefix_and_encode(dst->encoding());
4080   emit_int8((unsigned char)0xD3);
4081   emit_int8((unsigned char)(0xE0 | encode));
4082 }
4083 
4084 void Assembler::shrl(Register dst, int imm8) {
4085   assert(isShiftCount(imm8), "illegal shift count");
4086   int encode = prefix_and_encode(dst->encoding());
4087   emit_int8((unsigned char)0xC1);
4088   emit_int8((unsigned char)(0xE8 | encode));
4089   emit_int8(imm8);
4090 }
4091 
4092 void Assembler::shrl(Register dst) {
4093   int encode = prefix_and_encode(dst->encoding());
4094   emit_int8((unsigned char)0xD3);
4095   emit_int8((unsigned char)(0xE8 | encode));
4096 }
4097 
4098 // copies a single word from [esi] to [edi]
4099 void Assembler::smovl() {
4100   emit_int8((unsigned char)0xA5);
4101 }
4102 
4103 void Assembler::sqrtsd(XMMRegister dst, XMMRegister src) {
4104   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4105   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4106   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
4107   emit_int8(0x51);
4108   emit_int8((unsigned char)(0xC0 | encode));
4109 }
4110 
4111 void Assembler::sqrtsd(XMMRegister dst, Address src) {
4112   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4113   InstructionMark im(this);
4114   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4115   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
4116   simd_prefix(dst, dst, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
4117   emit_int8(0x51);
4118   emit_operand(dst, src);
4119 }
4120 
4121 void Assembler::sqrtss(XMMRegister dst, XMMRegister src) {
4122   NOT_LP64(assert(VM_Version::supports_sse(), ""));
4123   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4124   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
4125   emit_int8(0x51);
4126   emit_int8((unsigned char)(0xC0 | encode));
4127 }
4128 
4129 void Assembler::std() {
4130   emit_int8((unsigned char)0xFD);
4131 }
4132 
4133 void Assembler::sqrtss(XMMRegister dst, Address src) {
4134   NOT_LP64(assert(VM_Version::supports_sse(), ""));
4135   InstructionMark im(this);
4136   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4137   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
4138   simd_prefix(dst, dst, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
4139   emit_int8(0x51);
4140   emit_operand(dst, src);
4141 }
4142 
4143 void Assembler::stmxcsr( Address dst) {
4144   NOT_LP64(assert(VM_Version::supports_sse(), ""));
4145   InstructionMark im(this);
4146   prefix(dst);
4147   emit_int8(0x0F);
4148   emit_int8((unsigned char)0xAE);
4149   emit_operand(as_Register(3), dst);
4150 }
4151 
4152 void Assembler::subl(Address dst, int32_t imm32) {
4153   InstructionMark im(this);
4154   prefix(dst);
4155   emit_arith_operand(0x81, rbp, dst, imm32);
4156 }
4157 
4158 void Assembler::subl(Address dst, Register src) {
4159   InstructionMark im(this);
4160   prefix(dst, src);
4161   emit_int8(0x29);
4162   emit_operand(src, dst);
4163 }
4164 
4165 void Assembler::subl(Register dst, int32_t imm32) {
4166   prefix(dst);
4167   emit_arith(0x81, 0xE8, dst, imm32);
4168 }
4169 
4170 // Force generation of a 4 byte immediate value even if it fits into 8bit
4171 void Assembler::subl_imm32(Register dst, int32_t imm32) {
4172   prefix(dst);
4173   emit_arith_imm32(0x81, 0xE8, dst, imm32);
4174 }
4175 
4176 void Assembler::subl(Register dst, Address src) {
4177   InstructionMark im(this);
4178   prefix(src, dst);
4179   emit_int8(0x2B);
4180   emit_operand(dst, src);
4181 }
4182 
4183 void Assembler::subl(Register dst, Register src) {
4184   (void) prefix_and_encode(dst->encoding(), src->encoding());
4185   emit_arith(0x2B, 0xC0, dst, src);
4186 }
4187 
4188 void Assembler::subsd(XMMRegister dst, XMMRegister src) {
4189   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4190   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4191   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
4192   emit_int8(0x5C);
4193   emit_int8((unsigned char)(0xC0 | encode));
4194 }
4195 
4196 void Assembler::subsd(XMMRegister dst, Address src) {
4197   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4198   InstructionMark im(this);
4199   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4200   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
4201   simd_prefix(dst, dst, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
4202   emit_int8(0x5C);
4203   emit_operand(dst, src);
4204 }
4205 
4206 void Assembler::subss(XMMRegister dst, XMMRegister src) {
4207   NOT_LP64(assert(VM_Version::supports_sse(), ""));
4208   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false , /* uses_vl */ false);
4209   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
4210   emit_int8(0x5C);
4211   emit_int8((unsigned char)(0xC0 | encode));
4212 }
4213 
4214 void Assembler::subss(XMMRegister dst, Address src) {
4215   NOT_LP64(assert(VM_Version::supports_sse(), ""));
4216   InstructionMark im(this);
4217   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4218   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
4219   simd_prefix(dst, dst, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
4220   emit_int8(0x5C);
4221   emit_operand(dst, src);
4222 }
4223 
4224 void Assembler::testb(Register dst, int imm8) {
4225   NOT_LP64(assert(dst->has_byte_register(), "must have byte register"));
4226   (void) prefix_and_encode(dst->encoding(), true);
4227   emit_arith_b(0xF6, 0xC0, dst, imm8);
4228 }
4229 
4230 void Assembler::testb(Address dst, int imm8) {
4231   InstructionMark im(this);
4232   prefix(dst);
4233   emit_int8((unsigned char)0xF6);
4234   emit_operand(rax, dst, 1);
4235   emit_int8(imm8);
4236 }
4237 
4238 void Assembler::testl(Register dst, int32_t imm32) {
4239   // not using emit_arith because test
4240   // doesn't support sign-extension of
4241   // 8bit operands
4242   int encode = dst->encoding();
4243   if (encode == 0) {
4244     emit_int8((unsigned char)0xA9);
4245   } else {
4246     encode = prefix_and_encode(encode);
4247     emit_int8((unsigned char)0xF7);
4248     emit_int8((unsigned char)(0xC0 | encode));
4249   }
4250   emit_int32(imm32);
4251 }
4252 
4253 void Assembler::testl(Register dst, Register src) {
4254   (void) prefix_and_encode(dst->encoding(), src->encoding());
4255   emit_arith(0x85, 0xC0, dst, src);
4256 }
4257 
4258 void Assembler::testl(Register dst, Address src) {
4259   InstructionMark im(this);
4260   prefix(src, dst);
4261   emit_int8((unsigned char)0x85);
4262   emit_operand(dst, src);
4263 }
4264 
4265 void Assembler::tzcntl(Register dst, Register src) {
4266   assert(VM_Version::supports_bmi1(), "tzcnt instruction not supported");
4267   emit_int8((unsigned char)0xF3);
4268   int encode = prefix_and_encode(dst->encoding(), src->encoding());
4269   emit_int8(0x0F);
4270   emit_int8((unsigned char)0xBC);
4271   emit_int8((unsigned char)0xC0 | encode);
4272 }
4273 
4274 void Assembler::tzcntq(Register dst, Register src) {
4275   assert(VM_Version::supports_bmi1(), "tzcnt instruction not supported");
4276   emit_int8((unsigned char)0xF3);
4277   int encode = prefixq_and_encode(dst->encoding(), src->encoding());
4278   emit_int8(0x0F);
4279   emit_int8((unsigned char)0xBC);
4280   emit_int8((unsigned char)(0xC0 | encode));
4281 }
4282 
4283 void Assembler::ucomisd(XMMRegister dst, Address src) {
4284   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4285   InstructionMark im(this);
4286   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
4287   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
4288   simd_prefix(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4289   emit_int8(0x2E);
4290   emit_operand(dst, src);
4291 }
4292 
4293 void Assembler::ucomisd(XMMRegister dst, XMMRegister src) {
4294   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4295   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
4296   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4297   emit_int8(0x2E);
4298   emit_int8((unsigned char)(0xC0 | encode));
4299 }
4300 
4301 void Assembler::ucomiss(XMMRegister dst, Address src) {
4302   NOT_LP64(assert(VM_Version::supports_sse(), ""));
4303   InstructionMark im(this);
4304   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
4305   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
4306   simd_prefix(dst, xnoreg, src, VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4307   emit_int8(0x2E);
4308   emit_operand(dst, src);
4309 }
4310 
4311 void Assembler::ucomiss(XMMRegister dst, XMMRegister src) {
4312   NOT_LP64(assert(VM_Version::supports_sse(), ""));
4313   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
4314   int encode = simd_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4315   emit_int8(0x2E);
4316   emit_int8((unsigned char)(0xC0 | encode));
4317 }
4318 
4319 void Assembler::xabort(int8_t imm8) {
4320   emit_int8((unsigned char)0xC6);
4321   emit_int8((unsigned char)0xF8);
4322   emit_int8((unsigned char)(imm8 & 0xFF));
4323 }
4324 
4325 void Assembler::xaddl(Address dst, Register src) {
4326   InstructionMark im(this);
4327   prefix(dst, src);
4328   emit_int8(0x0F);
4329   emit_int8((unsigned char)0xC1);
4330   emit_operand(src, dst);
4331 }
4332 
4333 void Assembler::xbegin(Label& abort, relocInfo::relocType rtype) {
4334   InstructionMark im(this);
4335   relocate(rtype);
4336   if (abort.is_bound()) {
4337     address entry = target(abort);
4338     assert(entry != NULL, "abort entry NULL");
4339     intptr_t offset = entry - pc();
4340     emit_int8((unsigned char)0xC7);
4341     emit_int8((unsigned char)0xF8);
4342     emit_int32(offset - 6); // 2 opcode + 4 address
4343   } else {
4344     abort.add_patch_at(code(), locator());
4345     emit_int8((unsigned char)0xC7);
4346     emit_int8((unsigned char)0xF8);
4347     emit_int32(0);
4348   }
4349 }
4350 
4351 void Assembler::xchgl(Register dst, Address src) { // xchg
4352   InstructionMark im(this);
4353   prefix(src, dst);
4354   emit_int8((unsigned char)0x87);
4355   emit_operand(dst, src);
4356 }
4357 
4358 void Assembler::xchgl(Register dst, Register src) {
4359   int encode = prefix_and_encode(dst->encoding(), src->encoding());
4360   emit_int8((unsigned char)0x87);
4361   emit_int8((unsigned char)(0xC0 | encode));
4362 }
4363 
4364 void Assembler::xend() {
4365   emit_int8((unsigned char)0x0F);
4366   emit_int8((unsigned char)0x01);
4367   emit_int8((unsigned char)0xD5);
4368 }
4369 
4370 void Assembler::xgetbv() {
4371   emit_int8(0x0F);
4372   emit_int8(0x01);
4373   emit_int8((unsigned char)0xD0);
4374 }
4375 
4376 void Assembler::xorl(Register dst, int32_t imm32) {
4377   prefix(dst);
4378   emit_arith(0x81, 0xF0, dst, imm32);
4379 }
4380 
4381 void Assembler::xorl(Register dst, Address src) {
4382   InstructionMark im(this);
4383   prefix(src, dst);
4384   emit_int8(0x33);
4385   emit_operand(dst, src);
4386 }
4387 
4388 void Assembler::xorl(Register dst, Register src) {
4389   (void) prefix_and_encode(dst->encoding(), src->encoding());
4390   emit_arith(0x33, 0xC0, dst, src);
4391 }
4392 
4393 void Assembler::xorb(Register dst, Address src) {
4394   InstructionMark im(this);
4395   prefix(src, dst);
4396   emit_int8(0x32);
4397   emit_operand(dst, src);
4398 }
4399 
4400 // AVX 3-operands scalar float-point arithmetic instructions
4401 
4402 void Assembler::vaddsd(XMMRegister dst, XMMRegister nds, Address src) {
4403   assert(VM_Version::supports_avx(), "");
4404   InstructionMark im(this);
4405   InstructionAttr attributes(AVX_128bit, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4406   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
4407   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4408   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
4409   emit_int8(0x58);
4410   emit_operand(dst, src);
4411 }
4412 
4413 void Assembler::vaddsd(XMMRegister dst, XMMRegister nds, XMMRegister src) {
4414   assert(VM_Version::supports_avx(), "");
4415   InstructionAttr attributes(AVX_128bit, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4416   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4417   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
4418   emit_int8(0x58);
4419   emit_int8((unsigned char)(0xC0 | encode));
4420 }
4421 
4422 void Assembler::vaddss(XMMRegister dst, XMMRegister nds, Address src) {
4423   assert(VM_Version::supports_avx(), "");
4424   InstructionMark im(this);
4425   InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4426   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
4427   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4428   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
4429   emit_int8(0x58);
4430   emit_operand(dst, src);
4431 }
4432 
4433 void Assembler::vaddss(XMMRegister dst, XMMRegister nds, XMMRegister src) {
4434   assert(VM_Version::supports_avx(), "");
4435   InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4436   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4437   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
4438   emit_int8(0x58);
4439   emit_int8((unsigned char)(0xC0 | encode));
4440 }
4441 
4442 void Assembler::vdivsd(XMMRegister dst, XMMRegister nds, Address src) {
4443   assert(VM_Version::supports_avx(), "");
4444   InstructionMark im(this);
4445   InstructionAttr attributes(AVX_128bit, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4446   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
4447   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4448   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
4449   emit_int8(0x5E);
4450   emit_operand(dst, src);
4451 }
4452 
4453 void Assembler::vdivsd(XMMRegister dst, XMMRegister nds, XMMRegister src) {
4454   assert(VM_Version::supports_avx(), "");
4455   InstructionAttr attributes(AVX_128bit, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4456   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4457   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
4458   emit_int8(0x5E);
4459   emit_int8((unsigned char)(0xC0 | encode));
4460 }
4461 
4462 void Assembler::vdivss(XMMRegister dst, XMMRegister nds, Address src) {
4463   assert(VM_Version::supports_avx(), "");
4464   InstructionMark im(this);
4465   InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4466   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
4467   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4468   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
4469   emit_int8(0x5E);
4470   emit_operand(dst, src);
4471 }
4472 
4473 void Assembler::vdivss(XMMRegister dst, XMMRegister nds, XMMRegister src) {
4474   assert(VM_Version::supports_avx(), "");
4475   InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4476   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4477   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
4478   emit_int8(0x5E);
4479   emit_int8((unsigned char)(0xC0 | encode));
4480 }
4481 
4482 void Assembler::vmulsd(XMMRegister dst, XMMRegister nds, Address src) {
4483   assert(VM_Version::supports_avx(), "");
4484   InstructionMark im(this);
4485   InstructionAttr attributes(AVX_128bit, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4486   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
4487   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4488   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
4489   emit_int8(0x59);
4490   emit_operand(dst, src);
4491 }
4492 
4493 void Assembler::vmulsd(XMMRegister dst, XMMRegister nds, XMMRegister src) {
4494   assert(VM_Version::supports_avx(), "");
4495   InstructionAttr attributes(AVX_128bit, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4496   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4497   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
4498   emit_int8(0x59);
4499   emit_int8((unsigned char)(0xC0 | encode));
4500 }
4501 
4502 void Assembler::vmulss(XMMRegister dst, XMMRegister nds, Address src) {
4503   assert(VM_Version::supports_avx(), "");
4504   InstructionMark im(this);
4505   InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4506   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
4507   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4508   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
4509   emit_int8(0x59);
4510   emit_operand(dst, src);
4511 }
4512 
4513 void Assembler::vmulss(XMMRegister dst, XMMRegister nds, XMMRegister src) {
4514   assert(VM_Version::supports_avx(), "");
4515   InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4516   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4517   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
4518   emit_int8(0x59);
4519   emit_int8((unsigned char)(0xC0 | encode));
4520 }
4521 
4522 void Assembler::vsubsd(XMMRegister dst, XMMRegister nds, Address src) {
4523   assert(VM_Version::supports_avx(), "");
4524   InstructionMark im(this);
4525   InstructionAttr attributes(AVX_128bit, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4526   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
4527   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4528   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
4529   emit_int8(0x5C);
4530   emit_operand(dst, src);
4531 }
4532 
4533 void Assembler::vsubsd(XMMRegister dst, XMMRegister nds, XMMRegister src) {
4534   assert(VM_Version::supports_avx(), "");
4535   InstructionAttr attributes(AVX_128bit, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4536   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4537   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
4538   emit_int8(0x5C);
4539   emit_int8((unsigned char)(0xC0 | encode));
4540 }
4541 
4542 void Assembler::vsubss(XMMRegister dst, XMMRegister nds, Address src) {
4543   assert(VM_Version::supports_avx(), "");
4544   InstructionMark im(this);
4545   InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4546   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
4547   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4548   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
4549   emit_int8(0x5C);
4550   emit_operand(dst, src);
4551 }
4552 
4553 void Assembler::vsubss(XMMRegister dst, XMMRegister nds, XMMRegister src) {
4554   assert(VM_Version::supports_avx(), "");
4555   InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
4556   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4557   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
4558   emit_int8(0x5C);
4559   emit_int8((unsigned char)(0xC0 | encode));
4560 }
4561 
4562 //====================VECTOR ARITHMETIC=====================================
4563 
4564 // Float-point vector arithmetic
4565 
4566 void Assembler::addpd(XMMRegister dst, XMMRegister src) {
4567   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4568   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4569   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4570   emit_int8(0x58);
4571   emit_int8((unsigned char)(0xC0 | encode));
4572 }
4573 
4574 void Assembler::addpd(XMMRegister dst, Address src) {
4575   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4576   InstructionMark im(this);
4577   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4578   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_64bit);
4579   simd_prefix(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4580   emit_int8(0x58);
4581   emit_operand(dst, src);
4582 }
4583 
4584 
4585 void Assembler::addps(XMMRegister dst, XMMRegister src) {
4586   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4587   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4588   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4589   emit_int8(0x58);
4590   emit_int8((unsigned char)(0xC0 | encode));
4591 }
4592 
4593 void Assembler::vaddpd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
4594   assert(VM_Version::supports_avx(), "");
4595   InstructionAttr attributes(vector_len, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4596   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4597   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4598   emit_int8(0x58);
4599   emit_int8((unsigned char)(0xC0 | encode));
4600 }
4601 
4602 void Assembler::vaddps(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
4603   assert(VM_Version::supports_avx(), "");
4604   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4605   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4606   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4607   emit_int8(0x58);
4608   emit_int8((unsigned char)(0xC0 | encode));
4609 }
4610 
4611 void Assembler::vaddpd(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
4612   assert(VM_Version::supports_avx(), "");
4613   InstructionMark im(this);
4614   InstructionAttr attributes(vector_len, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4615   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_64bit);
4616   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4617   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4618   emit_int8(0x58);
4619   emit_operand(dst, src);
4620 }
4621 
4622 void Assembler::vaddps(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
4623   assert(VM_Version::supports_avx(), "");
4624   InstructionMark im(this);
4625   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4626   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_32bit);
4627   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4628   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4629   emit_int8(0x58);
4630   emit_operand(dst, src);
4631 }
4632 
4633 void Assembler::subpd(XMMRegister dst, XMMRegister src) {
4634   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4635   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4636   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4637   emit_int8(0x5C);
4638   emit_int8((unsigned char)(0xC0 | encode));
4639 }
4640 
4641 void Assembler::subps(XMMRegister dst, XMMRegister src) {
4642   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4643   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4644   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4645   emit_int8(0x5C);
4646   emit_int8((unsigned char)(0xC0 | encode));
4647 }
4648 
4649 void Assembler::vsubpd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
4650   assert(VM_Version::supports_avx(), "");
4651   InstructionAttr attributes(vector_len, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4652   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4653   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4654   emit_int8(0x5C);
4655   emit_int8((unsigned char)(0xC0 | encode));
4656 }
4657 
4658 void Assembler::vsubps(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
4659   assert(VM_Version::supports_avx(), "");
4660   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4661   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4662   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4663   emit_int8(0x5C);
4664   emit_int8((unsigned char)(0xC0 | encode));
4665 }
4666 
4667 void Assembler::vsubpd(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
4668   assert(VM_Version::supports_avx(), "");
4669   InstructionMark im(this);
4670   InstructionAttr attributes(vector_len, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4671   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_64bit);
4672   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4673   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4674   emit_int8(0x5C);
4675   emit_operand(dst, src);
4676 }
4677 
4678 void Assembler::vsubps(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
4679   assert(VM_Version::supports_avx(), "");
4680   InstructionMark im(this);
4681   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4682   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_32bit);
4683   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4684   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4685   emit_int8(0x5C);
4686   emit_operand(dst, src);
4687 }
4688 
4689 void Assembler::mulpd(XMMRegister dst, XMMRegister src) {
4690   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4691   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4692   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4693   emit_int8(0x59);
4694   emit_int8((unsigned char)(0xC0 | encode));
4695 }
4696 
4697 void Assembler::mulpd(XMMRegister dst, Address src) {
4698   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4699   InstructionMark im(this);
4700   InstructionAttr attributes(AVX_128bit, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4701   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_64bit);
4702   simd_prefix(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4703   emit_int8(0x59);
4704   emit_operand(dst, src);
4705 }
4706 
4707 void Assembler::mulps(XMMRegister dst, XMMRegister src) {
4708   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4709   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4710   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4711   emit_int8(0x59);
4712   emit_int8((unsigned char)(0xC0 | encode));
4713 }
4714 
4715 void Assembler::vmulpd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
4716   assert(VM_Version::supports_avx(), "");
4717   InstructionAttr attributes(vector_len, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4718   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4719   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4720   emit_int8(0x59);
4721   emit_int8((unsigned char)(0xC0 | encode));
4722 }
4723 
4724 void Assembler::vmulps(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
4725   assert(VM_Version::supports_avx(), "");
4726   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4727   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4728   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4729   emit_int8(0x59);
4730   emit_int8((unsigned char)(0xC0 | encode));
4731 }
4732 
4733 void Assembler::vmulpd(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
4734   assert(VM_Version::supports_avx(), "");
4735   InstructionMark im(this);
4736   InstructionAttr attributes(vector_len, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4737   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_64bit);
4738   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4739   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4740   emit_int8(0x59);
4741   emit_operand(dst, src);
4742 }
4743 
4744 void Assembler::vmulps(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
4745   assert(VM_Version::supports_avx(), "");
4746   InstructionMark im(this);
4747   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4748   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_32bit);
4749   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4750   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4751   emit_int8(0x59);
4752   emit_operand(dst, src);
4753 }
4754 
4755 void Assembler::divpd(XMMRegister dst, XMMRegister src) {
4756   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4757   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4758   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4759   emit_int8(0x5E);
4760   emit_int8((unsigned char)(0xC0 | encode));
4761 }
4762 
4763 void Assembler::divps(XMMRegister dst, XMMRegister src) {
4764   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4765   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4766   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4767   emit_int8(0x5E);
4768   emit_int8((unsigned char)(0xC0 | encode));
4769 }
4770 
4771 void Assembler::vdivpd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
4772   assert(VM_Version::supports_avx(), "");
4773   InstructionAttr attributes(vector_len, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4774   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4775   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4776   emit_int8(0x5E);
4777   emit_int8((unsigned char)(0xC0 | encode));
4778 }
4779 
4780 void Assembler::vdivps(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
4781   assert(VM_Version::supports_avx(), "");
4782   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4783   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4784   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4785   emit_int8(0x5E);
4786   emit_int8((unsigned char)(0xC0 | encode));
4787 }
4788 
4789 void Assembler::vdivpd(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
4790   assert(VM_Version::supports_avx(), "");
4791   InstructionMark im(this);
4792   InstructionAttr attributes(vector_len, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4793   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_64bit);
4794   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4795   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4796   emit_int8(0x5E);
4797   emit_operand(dst, src);
4798 }
4799 
4800 void Assembler::vdivps(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
4801   assert(VM_Version::supports_avx(), "");
4802   InstructionMark im(this);
4803   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4804   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_32bit);
4805   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4806   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4807   emit_int8(0x5E);
4808   emit_operand(dst, src);
4809 }
4810 
4811 void Assembler::vsqrtpd(XMMRegister dst, XMMRegister src, int vector_len) {
4812   assert(VM_Version::supports_avx(), "");
4813   InstructionAttr attributes(vector_len, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4814   int nds_enc = 0;
4815   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4816   emit_int8(0x51);
4817   emit_int8((unsigned char)(0xC0 | encode));
4818 }
4819 
4820 void Assembler::vsqrtpd(XMMRegister dst, Address src, int vector_len) {
4821   assert(VM_Version::supports_avx(), "");
4822   InstructionMark im(this);
4823   InstructionAttr attributes(vector_len, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4824   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_64bit);
4825   vex_prefix(src, 0, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4826   emit_int8(0x51);
4827   emit_operand(dst, src);
4828 }
4829 
4830 void Assembler::andpd(XMMRegister dst, XMMRegister src) {
4831   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4832   InstructionAttr attributes(AVX_128bit, /* rex_w */ !_legacy_mode_dq, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ false, /* uses_vl */ true);
4833   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4834   emit_int8(0x54);
4835   emit_int8((unsigned char)(0xC0 | encode));
4836 }
4837 
4838 void Assembler::andps(XMMRegister dst, XMMRegister src) {
4839   NOT_LP64(assert(VM_Version::supports_sse(), ""));
4840   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ false, /* uses_vl */ true);
4841   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4842   emit_int8(0x54);
4843   emit_int8((unsigned char)(0xC0 | encode));
4844 }
4845 
4846 void Assembler::andps(XMMRegister dst, Address src) {
4847   NOT_LP64(assert(VM_Version::supports_sse(), ""));
4848   InstructionMark im(this);
4849   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ false, /* uses_vl */ true);
4850   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_32bit);
4851   simd_prefix(dst, dst, src, VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4852   emit_int8(0x54);
4853   emit_operand(dst, src);
4854 }
4855 
4856 void Assembler::andpd(XMMRegister dst, Address src) {
4857   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4858   InstructionMark im(this);
4859   InstructionAttr attributes(AVX_128bit, /* rex_w */ !_legacy_mode_dq, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ false, /* uses_vl */ true);
4860   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_64bit);
4861   simd_prefix(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4862   emit_int8(0x54);
4863   emit_operand(dst, src);
4864 }
4865 
4866 void Assembler::vandpd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
4867   assert(VM_Version::supports_avx(), "");
4868   InstructionAttr attributes(vector_len, /* vex_w */ !_legacy_mode_dq, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ false, /* uses_vl */ true);
4869   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4870   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4871   emit_int8(0x54);
4872   emit_int8((unsigned char)(0xC0 | encode));
4873 }
4874 
4875 void Assembler::vandps(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
4876   assert(VM_Version::supports_avx(), "");
4877   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ false, /* uses_vl */ true);
4878   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4879   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4880   emit_int8(0x54);
4881   emit_int8((unsigned char)(0xC0 | encode));
4882 }
4883 
4884 void Assembler::vandpd(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
4885   assert(VM_Version::supports_avx(), "");
4886   InstructionMark im(this);
4887   InstructionAttr attributes(vector_len, /* vex_w */ !_legacy_mode_dq, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ false, /* uses_vl */ true);
4888   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_64bit);
4889   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4890   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4891   emit_int8(0x54);
4892   emit_operand(dst, src);
4893 }
4894 
4895 void Assembler::vandps(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
4896   assert(VM_Version::supports_avx(), "");
4897   InstructionMark im(this);
4898   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ false, /* uses_vl */ true);
4899   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_32bit);
4900   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4901   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4902   emit_int8(0x54);
4903   emit_operand(dst, src);
4904 }
4905 
4906 void Assembler::unpckhpd(XMMRegister dst, XMMRegister src) {
4907   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4908   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4909   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4910   emit_int8(0x15);
4911   emit_int8((unsigned char)(0xC0 | encode));
4912 }
4913 
4914 void Assembler::unpcklpd(XMMRegister dst, XMMRegister src) {
4915   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4916   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
4917   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4918   emit_int8(0x14);
4919   emit_int8((unsigned char)(0xC0 | encode));
4920 }
4921 
4922 void Assembler::xorpd(XMMRegister dst, XMMRegister src) {
4923   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4924   InstructionAttr attributes(AVX_128bit, /* rex_w */ !_legacy_mode_dq, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ false, /* uses_vl */ true);
4925   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4926   emit_int8(0x57);
4927   emit_int8((unsigned char)(0xC0 | encode));
4928 }
4929 
4930 void Assembler::xorps(XMMRegister dst, XMMRegister src) {
4931   NOT_LP64(assert(VM_Version::supports_sse(), ""));
4932   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ false, /* uses_vl */ true);
4933   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4934   emit_int8(0x57);
4935   emit_int8((unsigned char)(0xC0 | encode));
4936 }
4937 
4938 void Assembler::xorpd(XMMRegister dst, Address src) {
4939   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
4940   InstructionMark im(this);
4941   InstructionAttr attributes(AVX_128bit, /* rex_w */ !_legacy_mode_dq, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ false, /* uses_vl */ true);
4942   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_64bit);
4943   simd_prefix(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4944   emit_int8(0x57);
4945   emit_operand(dst, src);
4946 }
4947 
4948 void Assembler::xorps(XMMRegister dst, Address src) {
4949   NOT_LP64(assert(VM_Version::supports_sse(), ""));
4950   InstructionMark im(this);
4951   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ false, /* uses_vl */ true);
4952   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_32bit);
4953   simd_prefix(dst, dst, src, VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4954   emit_int8(0x57);
4955   emit_operand(dst, src);
4956 }
4957 
4958 void Assembler::vxorpd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
4959   assert(VM_Version::supports_avx(), "");
4960   InstructionAttr attributes(vector_len, /* vex_w */ !_legacy_mode_dq, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ false, /* uses_vl */ true);
4961   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4962   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4963   emit_int8(0x57);
4964   emit_int8((unsigned char)(0xC0 | encode));
4965 }
4966 
4967 void Assembler::vxorps(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
4968   assert(VM_Version::supports_avx(), "");
4969   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ false, /* uses_vl */ true);
4970   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4971   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4972   emit_int8(0x57);
4973   emit_int8((unsigned char)(0xC0 | encode));
4974 }
4975 
4976 void Assembler::vxorpd(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
4977   assert(VM_Version::supports_avx(), "");
4978   InstructionMark im(this);
4979   InstructionAttr attributes(vector_len, /* vex_w */ !_legacy_mode_dq, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ false, /* uses_vl */ true);
4980   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_64bit);
4981   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4982   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
4983   emit_int8(0x57);
4984   emit_operand(dst, src);
4985 }
4986 
4987 void Assembler::vxorps(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
4988   assert(VM_Version::supports_avx(), "");
4989   InstructionMark im(this);
4990   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ false, /* uses_vl */ true);
4991   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_32bit);
4992   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
4993   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
4994   emit_int8(0x57);
4995   emit_operand(dst, src);
4996 }
4997 
4998 // Integer vector arithmetic
4999 void Assembler::vphaddw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
5000   assert(VM_Version::supports_avx() && (vector_len == 0) ||
5001          VM_Version::supports_avx2(), "256 bit integer vectors requires AVX2");
5002   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
5003   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5004   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
5005   emit_int8(0x01);
5006   emit_int8((unsigned char)(0xC0 | encode));
5007 }
5008 
5009 void Assembler::vphaddd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
5010   assert(VM_Version::supports_avx() && (vector_len == 0) ||
5011          VM_Version::supports_avx2(), "256 bit integer vectors requires AVX2");
5012   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
5013   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5014   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
5015   emit_int8(0x02);
5016   emit_int8((unsigned char)(0xC0 | encode));
5017 }
5018 
5019 void Assembler::paddb(XMMRegister dst, XMMRegister src) {
5020   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5021   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5022   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5023   emit_int8((unsigned char)0xFC);
5024   emit_int8((unsigned char)(0xC0 | encode));
5025 }
5026 
5027 void Assembler::paddw(XMMRegister dst, XMMRegister src) {
5028   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5029   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5030   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5031   emit_int8((unsigned char)0xFD);
5032   emit_int8((unsigned char)(0xC0 | encode));
5033 }
5034 
5035 void Assembler::paddd(XMMRegister dst, XMMRegister src) {
5036   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5037   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5038   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5039   emit_int8((unsigned char)0xFE);
5040   emit_int8((unsigned char)(0xC0 | encode));
5041 }
5042 
5043 void Assembler::paddd(XMMRegister dst, Address src) {
5044   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5045   InstructionMark im(this);
5046   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5047   simd_prefix(dst, xnoreg, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5048   emit_int8((unsigned char)0xFE);
5049   emit_operand(dst, src);
5050 }
5051 
5052 void Assembler::paddq(XMMRegister dst, XMMRegister src) {
5053   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5054   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5055   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5056   emit_int8((unsigned char)0xD4);
5057   emit_int8((unsigned char)(0xC0 | encode));
5058 }
5059 
5060 void Assembler::phaddw(XMMRegister dst, XMMRegister src) {
5061   assert(VM_Version::supports_sse3(), "");
5062   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
5063   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
5064   emit_int8(0x01);
5065   emit_int8((unsigned char)(0xC0 | encode));
5066 }
5067 
5068 void Assembler::phaddd(XMMRegister dst, XMMRegister src) {
5069   assert(VM_Version::supports_sse3(), "");
5070   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
5071   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
5072   emit_int8(0x02);
5073   emit_int8((unsigned char)(0xC0 | encode));
5074 }
5075 
5076 void Assembler::vpaddb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
5077   assert(UseAVX > 0, "requires some form of AVX");
5078   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5079   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5080   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5081   emit_int8((unsigned char)0xFC);
5082   emit_int8((unsigned char)(0xC0 | encode));
5083 }
5084 
5085 void Assembler::vpaddw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
5086   assert(UseAVX > 0, "requires some form of AVX");
5087   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5088   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5089   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5090   emit_int8((unsigned char)0xFD);
5091   emit_int8((unsigned char)(0xC0 | encode));
5092 }
5093 
5094 void Assembler::vpaddd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
5095   assert(UseAVX > 0, "requires some form of AVX");
5096   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5097   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5098   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5099   emit_int8((unsigned char)0xFE);
5100   emit_int8((unsigned char)(0xC0 | encode));
5101 }
5102 
5103 void Assembler::vpaddq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
5104   assert(UseAVX > 0, "requires some form of AVX");
5105   InstructionAttr attributes(vector_len, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5106   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5107   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5108   emit_int8((unsigned char)0xD4);
5109   emit_int8((unsigned char)(0xC0 | encode));
5110 }
5111 
5112 void Assembler::vpaddb(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
5113   assert(UseAVX > 0, "requires some form of AVX");
5114   InstructionMark im(this);
5115   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5116   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
5117   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5118   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5119   emit_int8((unsigned char)0xFC);
5120   emit_operand(dst, src);
5121 }
5122 
5123 void Assembler::vpaddw(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
5124   assert(UseAVX > 0, "requires some form of AVX");
5125   InstructionMark im(this);
5126   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5127   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
5128   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5129   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5130   emit_int8((unsigned char)0xFD);
5131   emit_operand(dst, src);
5132 }
5133 
5134 void Assembler::vpaddd(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
5135   assert(UseAVX > 0, "requires some form of AVX");
5136   InstructionMark im(this);
5137   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5138   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_32bit);
5139   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5140   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5141   emit_int8((unsigned char)0xFE);
5142   emit_operand(dst, src);
5143 }
5144 
5145 void Assembler::vpaddq(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
5146   assert(UseAVX > 0, "requires some form of AVX");
5147   InstructionMark im(this);
5148   InstructionAttr attributes(vector_len, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5149   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_64bit);
5150   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5151   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5152   emit_int8((unsigned char)0xD4);
5153   emit_operand(dst, src);
5154 }
5155 
5156 void Assembler::psubb(XMMRegister dst, XMMRegister src) {
5157   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5158   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5159   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5160   emit_int8((unsigned char)0xF8);
5161   emit_int8((unsigned char)(0xC0 | encode));
5162 }
5163 
5164 void Assembler::psubw(XMMRegister dst, XMMRegister src) {
5165   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5166   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5167   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5168   emit_int8((unsigned char)0xF9);
5169   emit_int8((unsigned char)(0xC0 | encode));
5170 }
5171 
5172 void Assembler::psubd(XMMRegister dst, XMMRegister src) {
5173   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5174   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5175   emit_int8((unsigned char)0xFA);
5176   emit_int8((unsigned char)(0xC0 | encode));
5177 }
5178 
5179 void Assembler::psubq(XMMRegister dst, XMMRegister src) {
5180   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5181   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5182   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5183   emit_int8((unsigned char)0xFB);
5184   emit_int8((unsigned char)(0xC0 | encode));
5185 }
5186 
5187 void Assembler::vpsubb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
5188   assert(UseAVX > 0, "requires some form of AVX");
5189   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5190   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5191   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5192   emit_int8((unsigned char)0xF8);
5193   emit_int8((unsigned char)(0xC0 | encode));
5194 }
5195 
5196 void Assembler::vpsubw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
5197   assert(UseAVX > 0, "requires some form of AVX");
5198   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5199   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5200   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5201   emit_int8((unsigned char)0xF9);
5202   emit_int8((unsigned char)(0xC0 | encode));
5203 }
5204 
5205 void Assembler::vpsubd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
5206   assert(UseAVX > 0, "requires some form of AVX");
5207   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5208   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5209   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5210   emit_int8((unsigned char)0xFA);
5211   emit_int8((unsigned char)(0xC0 | encode));
5212 }
5213 
5214 void Assembler::vpsubq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
5215   assert(UseAVX > 0, "requires some form of AVX");
5216   InstructionAttr attributes(vector_len, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5217   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5218   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5219   emit_int8((unsigned char)0xFB);
5220   emit_int8((unsigned char)(0xC0 | encode));
5221 }
5222 
5223 void Assembler::vpsubb(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
5224   assert(UseAVX > 0, "requires some form of AVX");
5225   InstructionMark im(this);
5226   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5227   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
5228   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5229   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5230   emit_int8((unsigned char)0xF8);
5231   emit_operand(dst, src);
5232 }
5233 
5234 void Assembler::vpsubw(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
5235   assert(UseAVX > 0, "requires some form of AVX");
5236   InstructionMark im(this);
5237   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5238   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
5239   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5240   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5241   emit_int8((unsigned char)0xF9);
5242   emit_operand(dst, src);
5243 }
5244 
5245 void Assembler::vpsubd(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
5246   assert(UseAVX > 0, "requires some form of AVX");
5247   InstructionMark im(this);
5248   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5249   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_32bit);
5250   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5251   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5252   emit_int8((unsigned char)0xFA);
5253   emit_operand(dst, src);
5254 }
5255 
5256 void Assembler::vpsubq(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
5257   assert(UseAVX > 0, "requires some form of AVX");
5258   InstructionMark im(this);
5259   InstructionAttr attributes(vector_len, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5260   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_64bit);
5261   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5262   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5263   emit_int8((unsigned char)0xFB);
5264   emit_operand(dst, src);
5265 }
5266 
5267 void Assembler::pmullw(XMMRegister dst, XMMRegister src) {
5268   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5269   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5270   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5271   emit_int8((unsigned char)0xD5);
5272   emit_int8((unsigned char)(0xC0 | encode));
5273 }
5274 
5275 void Assembler::pmulld(XMMRegister dst, XMMRegister src) {
5276   assert(VM_Version::supports_sse4_1(), "");
5277   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5278   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
5279   emit_int8(0x40);
5280   emit_int8((unsigned char)(0xC0 | encode));
5281 }
5282 
5283 void Assembler::vpmullw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
5284   assert(UseAVX > 0, "requires some form of AVX");
5285   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5286   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5287   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5288   emit_int8((unsigned char)0xD5);
5289   emit_int8((unsigned char)(0xC0 | encode));
5290 }
5291 
5292 void Assembler::vpmulld(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
5293   assert(UseAVX > 0, "requires some form of AVX");
5294   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5295   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5296   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
5297   emit_int8(0x40);
5298   emit_int8((unsigned char)(0xC0 | encode));
5299 }
5300 
5301 void Assembler::vpmullq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
5302   assert(UseAVX > 2, "requires some form of AVX");
5303   InstructionAttr attributes(vector_len, /* rex_w */ true, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ false, /* uses_vl */ true);
5304   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5305   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
5306   emit_int8(0x40);
5307   emit_int8((unsigned char)(0xC0 | encode));
5308 }
5309 
5310 void Assembler::vpmullw(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
5311   assert(UseAVX > 0, "requires some form of AVX");
5312   InstructionMark im(this);
5313   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5314   attributes.set_address_attributes(/* tuple_type */ EVEX_FVM, /* input_size_in_bits */ EVEX_NObit);
5315   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5316   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5317   emit_int8((unsigned char)0xD5);
5318   emit_operand(dst, src);
5319 }
5320 
5321 void Assembler::vpmulld(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
5322   assert(UseAVX > 0, "requires some form of AVX");
5323   InstructionMark im(this);
5324   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5325   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_32bit);
5326   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5327   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
5328   emit_int8(0x40);
5329   emit_operand(dst, src);
5330 }
5331 
5332 void Assembler::vpmullq(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
5333   assert(UseAVX > 0, "requires some form of AVX");
5334   InstructionMark im(this);
5335   InstructionAttr attributes(vector_len, /* vex_w */ true, /* legacy_mode */ _legacy_mode_dq, /* no_mask_reg */ false, /* uses_vl */ true);
5336   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_64bit);
5337   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5338   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
5339   emit_int8(0x40);
5340   emit_operand(dst, src);
5341 }
5342 
5343 // Shift packed integers left by specified number of bits.
5344 void Assembler::psllw(XMMRegister dst, int shift) {
5345   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5346   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5347   // XMM6 is for /6 encoding: 66 0F 71 /6 ib
5348   int encode = simd_prefix_and_encode(xmm6, dst, dst, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5349   emit_int8(0x71);
5350   emit_int8((unsigned char)(0xC0 | encode));
5351   emit_int8(shift & 0xFF);
5352 }
5353 
5354 void Assembler::pslld(XMMRegister dst, int shift) {
5355   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5356   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5357   // XMM6 is for /6 encoding: 66 0F 72 /6 ib
5358   int encode = simd_prefix_and_encode(xmm6, dst, dst, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5359   emit_int8(0x72);
5360   emit_int8((unsigned char)(0xC0 | encode));
5361   emit_int8(shift & 0xFF);
5362 }
5363 
5364 void Assembler::psllq(XMMRegister dst, int shift) {
5365   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5366   InstructionAttr attributes(AVX_128bit, /* rex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5367   // XMM6 is for /6 encoding: 66 0F 73 /6 ib
5368   int encode = simd_prefix_and_encode(xmm6, dst, dst, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5369   emit_int8(0x73);
5370   emit_int8((unsigned char)(0xC0 | encode));
5371   emit_int8(shift & 0xFF);
5372 }
5373 
5374 void Assembler::psllw(XMMRegister dst, XMMRegister shift) {
5375   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5376   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5377   int encode = simd_prefix_and_encode(dst, dst, shift, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5378   emit_int8((unsigned char)0xF1);
5379   emit_int8((unsigned char)(0xC0 | encode));
5380 }
5381 
5382 void Assembler::pslld(XMMRegister dst, XMMRegister shift) {
5383   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5384   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5385   int encode = simd_prefix_and_encode(dst, dst, shift, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5386   emit_int8((unsigned char)0xF2);
5387   emit_int8((unsigned char)(0xC0 | encode));
5388 }
5389 
5390 void Assembler::psllq(XMMRegister dst, XMMRegister shift) {
5391   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5392   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5393   int encode = simd_prefix_and_encode(dst, dst, shift, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5394   emit_int8((unsigned char)0xF3);
5395   emit_int8((unsigned char)(0xC0 | encode));
5396 }
5397 
5398 void Assembler::vpsllw(XMMRegister dst, XMMRegister src, int shift, int vector_len) {
5399   assert(UseAVX > 0, "requires some form of AVX");
5400   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5401   // XMM6 is for /6 encoding: 66 0F 71 /6 ib
5402   int encode = vex_prefix_and_encode(xmm6->encoding(), dst->encoding(), src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5403   emit_int8(0x71);
5404   emit_int8((unsigned char)(0xC0 | encode));
5405   emit_int8(shift & 0xFF);
5406 }
5407 
5408 void Assembler::vpslld(XMMRegister dst, XMMRegister src, int shift, int vector_len) {
5409   assert(UseAVX > 0, "requires some form of AVX");
5410   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5411   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5412   // XMM6 is for /6 encoding: 66 0F 72 /6 ib
5413   int encode = vex_prefix_and_encode(xmm6->encoding(), dst->encoding(), src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5414   emit_int8(0x72);
5415   emit_int8((unsigned char)(0xC0 | encode));
5416   emit_int8(shift & 0xFF);
5417 }
5418 
5419 void Assembler::vpsllq(XMMRegister dst, XMMRegister src, int shift, int vector_len) {
5420   assert(UseAVX > 0, "requires some form of AVX");
5421   InstructionAttr attributes(vector_len, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5422   // XMM6 is for /6 encoding: 66 0F 73 /6 ib
5423   int encode = vex_prefix_and_encode(xmm6->encoding(), dst->encoding(), src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5424   emit_int8(0x73);
5425   emit_int8((unsigned char)(0xC0 | encode));
5426   emit_int8(shift & 0xFF);
5427 }
5428 
5429 void Assembler::vpsllw(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len) {
5430   assert(UseAVX > 0, "requires some form of AVX");
5431   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5432   int encode = vex_prefix_and_encode(dst->encoding(), src->encoding(), shift->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5433   emit_int8((unsigned char)0xF1);
5434   emit_int8((unsigned char)(0xC0 | encode));
5435 }
5436 
5437 void Assembler::vpslld(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len) {
5438   assert(UseAVX > 0, "requires some form of AVX");
5439   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5440   int encode = vex_prefix_and_encode(dst->encoding(), src->encoding(), shift->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5441   emit_int8((unsigned char)0xF2);
5442   emit_int8((unsigned char)(0xC0 | encode));
5443 }
5444 
5445 void Assembler::vpsllq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len) {
5446   assert(UseAVX > 0, "requires some form of AVX");
5447   InstructionAttr attributes(vector_len, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5448   int encode = vex_prefix_and_encode(dst->encoding(), src->encoding(), shift->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5449   emit_int8((unsigned char)0xF3);
5450   emit_int8((unsigned char)(0xC0 | encode));
5451 }
5452 
5453 // Shift packed integers logically right by specified number of bits.
5454 void Assembler::psrlw(XMMRegister dst, int shift) {
5455   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5456   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5457   // XMM2 is for /2 encoding: 66 0F 71 /2 ib
5458   int encode = simd_prefix_and_encode(xmm2, dst, dst, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5459   emit_int8(0x71);
5460   emit_int8((unsigned char)(0xC0 | encode));
5461   emit_int8(shift & 0xFF);
5462 }
5463 
5464 void Assembler::psrld(XMMRegister dst, int shift) {
5465   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5466   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5467   // XMM2 is for /2 encoding: 66 0F 72 /2 ib
5468   int encode = simd_prefix_and_encode(xmm2, dst, dst, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5469   emit_int8(0x72);
5470   emit_int8((unsigned char)(0xC0 | encode));
5471   emit_int8(shift & 0xFF);
5472 }
5473 
5474 void Assembler::psrlq(XMMRegister dst, int shift) {
5475   // Do not confuse it with psrldq SSE2 instruction which
5476   // shifts 128 bit value in xmm register by number of bytes.
5477   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5478   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5479   // XMM2 is for /2 encoding: 66 0F 73 /2 ib
5480   int encode = simd_prefix_and_encode(xmm2, dst, dst, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5481   emit_int8(0x73);
5482   emit_int8((unsigned char)(0xC0 | encode));
5483   emit_int8(shift & 0xFF);
5484 }
5485 
5486 void Assembler::psrlw(XMMRegister dst, XMMRegister shift) {
5487   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5488   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5489   int encode = simd_prefix_and_encode(dst, dst, shift, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5490   emit_int8((unsigned char)0xD1);
5491   emit_int8((unsigned char)(0xC0 | encode));
5492 }
5493 
5494 void Assembler::psrld(XMMRegister dst, XMMRegister shift) {
5495   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5496   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5497   int encode = simd_prefix_and_encode(dst, dst, shift, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5498   emit_int8((unsigned char)0xD2);
5499   emit_int8((unsigned char)(0xC0 | encode));
5500 }
5501 
5502 void Assembler::psrlq(XMMRegister dst, XMMRegister shift) {
5503   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5504   InstructionAttr attributes(AVX_128bit, /* rex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5505   int encode = simd_prefix_and_encode(dst, dst, shift, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5506   emit_int8((unsigned char)0xD3);
5507   emit_int8((unsigned char)(0xC0 | encode));
5508 }
5509 
5510 void Assembler::vpsrlw(XMMRegister dst, XMMRegister src, int shift, int vector_len) {
5511   assert(UseAVX > 0, "requires some form of AVX");
5512   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5513   // XMM2 is for /2 encoding: 66 0F 71 /2 ib
5514   int encode = vex_prefix_and_encode(xmm2->encoding(), dst->encoding(), src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5515   emit_int8(0x71);
5516   emit_int8((unsigned char)(0xC0 | encode));
5517   emit_int8(shift & 0xFF);
5518 }
5519 
5520 void Assembler::vpsrld(XMMRegister dst, XMMRegister src, int shift, int vector_len) {
5521   assert(UseAVX > 0, "requires some form of AVX");
5522   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5523   // XMM2 is for /2 encoding: 66 0F 72 /2 ib
5524   int encode = vex_prefix_and_encode(xmm2->encoding(), dst->encoding(), src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5525   emit_int8(0x72);
5526   emit_int8((unsigned char)(0xC0 | encode));
5527   emit_int8(shift & 0xFF);
5528 }
5529 
5530 void Assembler::vpsrlq(XMMRegister dst, XMMRegister src, int shift, int vector_len) {
5531   assert(UseAVX > 0, "requires some form of AVX");
5532   InstructionAttr attributes(vector_len, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5533   // XMM2 is for /2 encoding: 66 0F 73 /2 ib
5534   int encode = vex_prefix_and_encode(xmm2->encoding(), dst->encoding(), src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5535   emit_int8(0x73);
5536   emit_int8((unsigned char)(0xC0 | encode));
5537   emit_int8(shift & 0xFF);
5538 }
5539 
5540 void Assembler::vpsrlw(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len) {
5541   assert(UseAVX > 0, "requires some form of AVX");
5542   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5543   int encode = vex_prefix_and_encode(dst->encoding(), src->encoding(), shift->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5544   emit_int8((unsigned char)0xD1);
5545   emit_int8((unsigned char)(0xC0 | encode));
5546 }
5547 
5548 void Assembler::vpsrld(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len) {
5549   assert(UseAVX > 0, "requires some form of AVX");
5550   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5551   int encode = vex_prefix_and_encode(dst->encoding(), src->encoding(), shift->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5552   emit_int8((unsigned char)0xD2);
5553   emit_int8((unsigned char)(0xC0 | encode));
5554 }
5555 
5556 void Assembler::vpsrlq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len) {
5557   assert(UseAVX > 0, "requires some form of AVX");
5558   InstructionAttr attributes(vector_len, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5559   int encode = vex_prefix_and_encode(dst->encoding(), src->encoding(), shift->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5560   emit_int8((unsigned char)0xD3);
5561   emit_int8((unsigned char)(0xC0 | encode));
5562 }
5563 
5564 // Shift packed integers arithmetically right by specified number of bits.
5565 void Assembler::psraw(XMMRegister dst, int shift) {
5566   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5567   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5568   // XMM4 is for /4 encoding: 66 0F 71 /4 ib
5569   int encode = simd_prefix_and_encode(xmm4, dst, dst, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5570   emit_int8(0x71);
5571   emit_int8((unsigned char)(0xC0 | encode));
5572   emit_int8(shift & 0xFF);
5573 }
5574 
5575 void Assembler::psrad(XMMRegister dst, int shift) {
5576   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5577   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5578   // XMM4 is for /4 encoding: 66 0F 72 /4 ib
5579   int encode = simd_prefix_and_encode(xmm4, dst, dst, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5580   emit_int8(0x72);
5581   emit_int8((unsigned char)(0xC0 | encode));
5582   emit_int8(shift & 0xFF);
5583 }
5584 
5585 void Assembler::psraw(XMMRegister dst, XMMRegister shift) {
5586   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5587   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5588   int encode = simd_prefix_and_encode(dst, dst, shift, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5589   emit_int8((unsigned char)0xE1);
5590   emit_int8((unsigned char)(0xC0 | encode));
5591 }
5592 
5593 void Assembler::psrad(XMMRegister dst, XMMRegister shift) {
5594   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5595   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5596   int encode = simd_prefix_and_encode(dst, dst, shift, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5597   emit_int8((unsigned char)0xE2);
5598   emit_int8((unsigned char)(0xC0 | encode));
5599 }
5600 
5601 void Assembler::vpsraw(XMMRegister dst, XMMRegister src, int shift, int vector_len) {
5602   assert(UseAVX > 0, "requires some form of AVX");
5603   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5604   // XMM4 is for /4 encoding: 66 0F 71 /4 ib
5605   int encode = vex_prefix_and_encode(xmm4->encoding(), dst->encoding(), src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5606   emit_int8(0x71);
5607   emit_int8((unsigned char)(0xC0 | encode));
5608   emit_int8(shift & 0xFF);
5609 }
5610 
5611 void Assembler::vpsrad(XMMRegister dst, XMMRegister src, int shift, int vector_len) {
5612   assert(UseAVX > 0, "requires some form of AVX");
5613   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5614   // XMM4 is for /4 encoding: 66 0F 71 /4 ib
5615   int encode = vex_prefix_and_encode(xmm4->encoding(), dst->encoding(), src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5616   emit_int8(0x72);
5617   emit_int8((unsigned char)(0xC0 | encode));
5618   emit_int8(shift & 0xFF);
5619 }
5620 
5621 void Assembler::vpsraw(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len) {
5622   assert(UseAVX > 0, "requires some form of AVX");
5623   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
5624   int encode = vex_prefix_and_encode(dst->encoding(), src->encoding(), shift->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5625   emit_int8((unsigned char)0xE1);
5626   emit_int8((unsigned char)(0xC0 | encode));
5627 }
5628 
5629 void Assembler::vpsrad(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len) {
5630   assert(UseAVX > 0, "requires some form of AVX");
5631   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5632   int encode = vex_prefix_and_encode(dst->encoding(), src->encoding(), shift->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5633   emit_int8((unsigned char)0xE2);
5634   emit_int8((unsigned char)(0xC0 | encode));
5635 }
5636 
5637 
5638 // logical operations packed integers
5639 void Assembler::pand(XMMRegister dst, XMMRegister src) {
5640   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5641   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5642   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5643   emit_int8((unsigned char)0xDB);
5644   emit_int8((unsigned char)(0xC0 | encode));
5645 }
5646 
5647 void Assembler::vpand(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
5648   assert(UseAVX > 0, "requires some form of AVX");
5649   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5650   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5651   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5652   emit_int8((unsigned char)0xDB);
5653   emit_int8((unsigned char)(0xC0 | encode));
5654 }
5655 
5656 void Assembler::vpand(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
5657   assert(UseAVX > 0, "requires some form of AVX");
5658   InstructionMark im(this);
5659   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5660   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_32bit);
5661   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5662   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5663   emit_int8((unsigned char)0xDB);
5664   emit_operand(dst, src);
5665 }
5666 
5667 void Assembler::pandn(XMMRegister dst, XMMRegister src) {
5668   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5669   InstructionAttr attributes(AVX_128bit, /* vex_w */ VM_Version::supports_evex(), /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5670   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5671   emit_int8((unsigned char)0xDF);
5672   emit_int8((unsigned char)(0xC0 | encode));
5673 }
5674 
5675 void Assembler::por(XMMRegister dst, XMMRegister src) {
5676   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5677   InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5678   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5679   emit_int8((unsigned char)0xEB);
5680   emit_int8((unsigned char)(0xC0 | encode));
5681 }
5682 
5683 void Assembler::vpor(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
5684   assert(UseAVX > 0, "requires some form of AVX");
5685   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5686   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5687   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5688   emit_int8((unsigned char)0xEB);
5689   emit_int8((unsigned char)(0xC0 | encode));
5690 }
5691 
5692 void Assembler::vpor(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
5693   assert(UseAVX > 0, "requires some form of AVX");
5694   InstructionMark im(this);
5695   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5696   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_32bit);
5697   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5698   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5699   emit_int8((unsigned char)0xEB);
5700   emit_operand(dst, src);
5701 }
5702 
5703 void Assembler::pxor(XMMRegister dst, XMMRegister src) {
5704   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
5705   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5706   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5707   emit_int8((unsigned char)0xEF);
5708   emit_int8((unsigned char)(0xC0 | encode));
5709 }
5710 
5711 void Assembler::vpxor(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len) {
5712   assert(UseAVX > 0, "requires some form of AVX");
5713   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5714   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5715   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5716   emit_int8((unsigned char)0xEF);
5717   emit_int8((unsigned char)(0xC0 | encode));
5718 }
5719 
5720 void Assembler::vpxor(XMMRegister dst, XMMRegister nds, Address src, int vector_len) {
5721   assert(UseAVX > 0, "requires some form of AVX");
5722   InstructionMark im(this);
5723   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
5724   attributes.set_address_attributes(/* tuple_type */ EVEX_FV, /* input_size_in_bits */ EVEX_32bit);
5725   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5726   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
5727   emit_int8((unsigned char)0xEF);
5728   emit_operand(dst, src);
5729 }
5730 
5731 
5732 void Assembler::vinsertf128(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8) {
5733   assert(VM_Version::supports_avx(), "");
5734   assert(imm8 <= 0x01, "imm8: %u", imm8);
5735   int vector_len = VM_Version::supports_avx512novl() ? AVX_512bit : AVX_256bit;
5736   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
5737   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5738   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
5739   emit_int8(0x18);
5740   emit_int8((unsigned char)(0xC0 | encode));
5741   // 0x00 - insert into lower 128 bits
5742   // 0x01 - insert into upper 128 bits
5743   emit_int8(imm8 & 0x01);
5744 }
5745 
5746 void Assembler::vinsertf64x4(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8) {
5747   assert(VM_Version::supports_evex(), "");
5748   assert(imm8 <= 0x01, "imm8: %u", imm8);
5749   InstructionAttr attributes(AVX_512bit, /* vex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
5750   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5751   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
5752   emit_int8(0x1A);
5753   emit_int8((unsigned char)(0xC0 | encode));
5754   // 0x00 - insert into lower 256 bits
5755   // 0x01 - insert into upper 256 bits
5756   emit_int8(imm8 & 0x01);
5757 }
5758 
5759 void Assembler::vinsertf64x4(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8) {
5760   assert(VM_Version::supports_evex(), "");
5761   assert(dst != xnoreg, "sanity");
5762   assert(imm8 <= 0x01, "imm8: %u", imm8);
5763   InstructionMark im(this);
5764   InstructionAttr attributes(AVX_512bit, /* vex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
5765   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5766   attributes.set_address_attributes(/* tuple_type */ EVEX_T4, /* input_size_in_bits */ EVEX_64bit);
5767   // swap src<->dst for encoding
5768   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
5769   emit_int8(0x1A);
5770   emit_operand(dst, src);
5771   // 0x00 - insert into lower 256 bits
5772   // 0x01 - insert into upper 256 bits
5773   emit_int8(imm8 & 0x01);
5774 }
5775 
5776 void Assembler::vinsertf32x4(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8) {
5777   assert(VM_Version::supports_evex(), "");
5778   assert(imm8 <= 0x03, "imm8: %u", imm8);
5779   InstructionAttr attributes(AVX_512bit, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
5780   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5781   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
5782   emit_int8(0x18);
5783   emit_int8((unsigned char)(0xC0 | encode));
5784   // 0x00 - insert into q0 128 bits (0..127)
5785   // 0x01 - insert into q1 128 bits (128..255)
5786   // 0x02 - insert into q2 128 bits (256..383)
5787   // 0x03 - insert into q3 128 bits (384..511)
5788   emit_int8(imm8 & 0x03);
5789 }
5790 
5791 void Assembler::vinsertf32x4(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8) {
5792   assert(VM_Version::supports_avx(), "");
5793   assert(dst != xnoreg, "sanity");
5794   assert(imm8 <= 0x03, "imm8: %u", imm8);
5795   int vector_len = VM_Version::supports_evex() ? AVX_512bit : AVX_256bit;
5796   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5797   InstructionMark im(this);
5798   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
5799   attributes.set_address_attributes(/* tuple_type */ EVEX_T4, /* input_size_in_bits */ EVEX_32bit);
5800   // swap src<->dst for encoding
5801   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
5802   emit_int8(0x18);
5803   emit_operand(dst, src);
5804   // 0x00 - insert into q0 128 bits (0..127)
5805   // 0x01 - insert into q1 128 bits (128..255)
5806   // 0x02 - insert into q2 128 bits (256..383)
5807   // 0x03 - insert into q3 128 bits (384..511)
5808   emit_int8(imm8 & 0x03);
5809 }
5810 
5811 void Assembler::vinsertf128(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8) {
5812   assert(VM_Version::supports_avx(), "");
5813   assert(dst != xnoreg, "sanity");
5814   assert(imm8 <= 0x01, "imm8: %u", imm8);
5815   int vector_len = VM_Version::supports_avx512novl() ? AVX_512bit : AVX_256bit;
5816   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5817   InstructionMark im(this);
5818   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
5819   attributes.set_address_attributes(/* tuple_type */ EVEX_T4, /* input_size_in_bits */ EVEX_32bit);
5820   // swap src<->dst for encoding
5821   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
5822   emit_int8(0x18);
5823   emit_operand(dst, src);
5824   // 0x00 - insert into lower 128 bits
5825   // 0x01 - insert into upper 128 bits
5826   emit_int8(imm8 & 0x01);
5827 }
5828 
5829 void Assembler::vextractf128(XMMRegister dst, XMMRegister src, uint8_t imm8) {
5830   assert(VM_Version::supports_avx(), "");
5831   assert(imm8 <= 0x01, "imm8: %u", imm8);
5832   int vector_len = VM_Version::supports_avx512novl() ? AVX_512bit : AVX_256bit;
5833   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
5834   int encode = vex_prefix_and_encode(src->encoding(), 0, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
5835   emit_int8(0x19);
5836   emit_int8((unsigned char)(0xC0 | encode));
5837   // 0x00 - extract from lower 128 bits
5838   // 0x01 - extract from upper 128 bits
5839   emit_int8(imm8 & 0x01);
5840 }
5841 
5842 void Assembler::vextractf128(Address dst, XMMRegister src, uint8_t imm8) {
5843   assert(VM_Version::supports_avx(), "");
5844   assert(src != xnoreg, "sanity");
5845   assert(imm8 <= 0x01, "imm8: %u", imm8);
5846   int vector_len = VM_Version::supports_avx512novl() ? AVX_512bit : AVX_256bit;
5847   InstructionMark im(this);
5848   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
5849   attributes.set_address_attributes(/* tuple_type */ EVEX_T4, /* input_size_in_bits */ EVEX_32bit);
5850   vex_prefix(dst, 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
5851   emit_int8(0x19);
5852   emit_operand(src, dst);
5853   // 0x00 - extract from lower 128 bits
5854   // 0x01 - extract from upper 128 bits
5855   emit_int8(imm8 & 0x01);
5856 }
5857 
5858 void Assembler::vinserti128(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8) {
5859   assert(VM_Version::supports_avx2(), "");
5860   assert(imm8 <= 0x01, "imm8: %u", imm8);
5861   int vector_len = VM_Version::supports_avx512novl() ? AVX_512bit : AVX_256bit;
5862   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
5863   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5864   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
5865   emit_int8(0x38);
5866   emit_int8((unsigned char)(0xC0 | encode));
5867   // 0x00 - insert into lower 128 bits
5868   // 0x01 - insert into upper 128 bits
5869   emit_int8(imm8 & 0x01);
5870 }
5871 
5872 void Assembler::vinserti64x4(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8) {
5873   assert(VM_Version::supports_evex(), "");
5874   assert(imm8 <= 0x01, "imm8: %u", imm8);
5875   InstructionAttr attributes(AVX_512bit, /* vex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
5876   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5877   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
5878   emit_int8(0x38);
5879   emit_int8((unsigned char)(0xC0 | encode));
5880   // 0x00 - insert into lower 256 bits
5881   // 0x01 - insert into upper 256 bits
5882   emit_int8(imm8 & 0x01);
5883 }
5884 
5885 void Assembler::vinserti128(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8) {
5886   assert(VM_Version::supports_avx2(), "");
5887   assert(dst != xnoreg, "sanity");
5888   assert(imm8 <= 0x01, "imm8: %u", imm8);
5889   int vector_len = VM_Version::supports_avx512novl() ? AVX_512bit : AVX_256bit;
5890   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
5891   InstructionMark im(this);
5892   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
5893   attributes.set_address_attributes(/* tuple_type */ EVEX_T4, /* input_size_in_bits */ EVEX_32bit);
5894   // swap src<->dst for encoding
5895   vex_prefix(src, nds_enc, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
5896   emit_int8(0x38);
5897   emit_operand(dst, src);
5898   // 0x00 - insert into lower 128 bits
5899   // 0x01 - insert into upper 128 bits
5900   emit_int8(imm8 & 0x01);
5901 }
5902 
5903 void Assembler::vextracti128(XMMRegister dst, XMMRegister src, uint8_t imm8) {
5904   assert(VM_Version::supports_avx(), "");
5905   assert(imm8 <= 0x01, "imm8: %u", imm8);
5906   int vector_len = VM_Version::supports_avx512novl() ? AVX_512bit : AVX_256bit;
5907   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
5908   int encode = vex_prefix_and_encode(src->encoding(), 0, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
5909   emit_int8(0x39);
5910   emit_int8((unsigned char)(0xC0 | encode));
5911   // 0x00 - extract from lower 128 bits
5912   // 0x01 - extract from upper 128 bits
5913   emit_int8(imm8 & 0x01);
5914 }
5915 
5916 void Assembler::vextracti128(Address dst, XMMRegister src, uint8_t imm8) {
5917   assert(VM_Version::supports_avx2(), "");
5918   assert(src != xnoreg, "sanity");
5919   assert(imm8 <= 0x01, "imm8: %u", imm8);
5920   int vector_len = VM_Version::supports_avx512novl() ? AVX_512bit : AVX_256bit;
5921   InstructionMark im(this);
5922   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
5923   attributes.set_address_attributes(/* tuple_type */ EVEX_T4, /* input_size_in_bits */ EVEX_32bit);
5924   vex_prefix(dst, 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
5925   emit_int8(0x39);
5926   emit_operand(src, dst);
5927   // 0x00 - extract from lower 128 bits
5928   // 0x01 - extract from upper 128 bits
5929   emit_int8(imm8 & 0x01);
5930 }
5931 
5932 void Assembler::vextracti64x4(XMMRegister dst, XMMRegister src, uint8_t imm8) {
5933   assert(VM_Version::supports_evex(), "");
5934   assert(imm8 <= 0x01, "imm8: %u", imm8);
5935   InstructionAttr attributes(AVX_512bit, /* vex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
5936   int encode = vex_prefix_and_encode(src->encoding(), 0, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
5937   emit_int8(0x3B);
5938   emit_int8((unsigned char)(0xC0 | encode));
5939   // 0x00 - extract from lower 256 bits
5940   // 0x01 - extract from upper 256 bits
5941   emit_int8(imm8 & 0x01);
5942 }
5943 
5944 void Assembler::vextracti64x2(XMMRegister dst, XMMRegister src, uint8_t imm8) {
5945   assert(VM_Version::supports_evex(), "");
5946   assert(imm8 <= 0x03, "imm8: %u", imm8);
5947   InstructionAttr attributes(AVX_512bit, /* vex_w */ !_legacy_mode_dq, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
5948   int encode = vex_prefix_and_encode(src->encoding(), 0, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
5949   emit_int8(0x39);
5950   emit_int8((unsigned char)(0xC0 | encode));
5951   // 0x00 - extract from bits 127:0
5952   // 0x01 - extract from bits 255:128
5953   // 0x02 - extract from bits 383:256
5954   // 0x03 - extract from bits 511:384
5955   emit_int8(imm8 & 0x03);
5956 }
5957 
5958 void Assembler::vextractf64x4(XMMRegister dst, XMMRegister src, uint8_t imm8) {
5959   assert(VM_Version::supports_evex(), "");
5960   assert(imm8 <= 0x01, "imm8: %u", imm8);
5961   InstructionAttr attributes(AVX_512bit, /* vex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
5962   int encode = vex_prefix_and_encode(src->encoding(), 0, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
5963   emit_int8(0x1B);
5964   emit_int8((unsigned char)(0xC0 | encode));
5965   // 0x00 - extract from lower 256 bits
5966   // 0x01 - extract from upper 256 bits
5967   emit_int8(imm8 & 0x01);
5968 }
5969 
5970 void Assembler::vextractf64x4(Address dst, XMMRegister src, uint8_t imm8) {
5971   assert(VM_Version::supports_evex(), "");
5972   assert(src != xnoreg, "sanity");
5973   assert(imm8 <= 0x01, "imm8: %u", imm8);
5974   InstructionMark im(this);
5975   InstructionAttr attributes(AVX_512bit, /* vex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
5976   attributes.set_address_attributes(/* tuple_type */ EVEX_T4,/* input_size_in_bits */  EVEX_64bit);
5977   vex_prefix(dst, 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
5978   emit_int8(0x1B);
5979   emit_operand(src, dst);
5980   // 0x00 - extract from lower 256 bits
5981   // 0x01 - extract from upper 256 bits
5982   emit_int8(imm8 & 0x01);
5983 }
5984 
5985 void Assembler::vextractf32x4(XMMRegister dst, XMMRegister src, uint8_t imm8) {
5986   assert(VM_Version::supports_avx(), "");
5987   assert(imm8 <= 0x03, "imm8: %u", imm8);
5988   int vector_len = VM_Version::supports_evex() ? AVX_512bit : AVX_256bit;
5989   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
5990   int encode = vex_prefix_and_encode(src->encoding(), 0, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
5991   emit_int8(0x19);
5992   emit_int8((unsigned char)(0xC0 | encode));
5993   // 0x00 - extract from bits 127:0
5994   // 0x01 - extract from bits 255:128
5995   // 0x02 - extract from bits 383:256
5996   // 0x03 - extract from bits 511:384
5997   emit_int8(imm8 & 0x03);
5998 }
5999 
6000 void Assembler::vextractf32x4(Address dst, XMMRegister src, uint8_t imm8) {
6001   assert(VM_Version::supports_evex(), "");
6002   assert(src != xnoreg, "sanity");
6003   assert(imm8 <= 0x03, "imm8: %u", imm8);
6004   InstructionMark im(this);
6005   InstructionAttr attributes(AVX_512bit, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
6006   attributes.set_address_attributes(/* tuple_type */ EVEX_T4, /* input_size_in_bits */ EVEX_32bit);
6007   vex_prefix(dst, 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
6008   emit_int8(0x19);
6009   emit_operand(src, dst);
6010   // 0x00 - extract from bits 127:0
6011   // 0x01 - extract from bits 255:128
6012   // 0x02 - extract from bits 383:256
6013   // 0x03 - extract from bits 511:384
6014   emit_int8(imm8 & 0x03);
6015 }
6016 
6017 void Assembler::vextractf64x2(XMMRegister dst, XMMRegister src, uint8_t imm8) {
6018   assert(VM_Version::supports_evex(), "");
6019   assert(imm8 <= 0x03, "imm8: %u", imm8);
6020   InstructionAttr attributes(AVX_512bit, /* vex_w */ !_legacy_mode_dq, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
6021   int encode = vex_prefix_and_encode(src->encoding(), 0, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
6022   emit_int8(0x19);
6023   emit_int8((unsigned char)(0xC0 | encode));
6024   // 0x00 - extract from bits 127:0
6025   // 0x01 - extract from bits 255:128
6026   // 0x02 - extract from bits 383:256
6027   // 0x03 - extract from bits 511:384
6028   emit_int8(imm8 & 0x03);
6029 }
6030 
6031 // duplicate 4-bytes integer data from src into 8 locations in dest
6032 void Assembler::vpbroadcastd(XMMRegister dst, XMMRegister src) {
6033   assert(VM_Version::supports_avx2(), "");
6034   InstructionAttr attributes(AVX_256bit, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
6035   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
6036   emit_int8(0x58);
6037   emit_int8((unsigned char)(0xC0 | encode));
6038 }
6039 
6040 // duplicate 2-bytes integer data from src into 16 locations in dest
6041 void Assembler::vpbroadcastw(XMMRegister dst, XMMRegister src) {
6042   assert(VM_Version::supports_avx2(), "");
6043   InstructionAttr attributes(AVX_256bit, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
6044   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
6045   emit_int8(0x79);
6046   emit_int8((unsigned char)(0xC0 | encode));
6047 }
6048 
6049 // duplicate 1-byte integer data from src into 16||32|64 locations in dest : requires AVX512BW and AVX512VL
6050 void Assembler::evpbroadcastb(XMMRegister dst, XMMRegister src, int vector_len) {
6051   assert(VM_Version::supports_evex(), "");
6052   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
6053   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
6054   emit_int8(0x78);
6055   emit_int8((unsigned char)(0xC0 | encode));
6056 }
6057 
6058 void Assembler::evpbroadcastb(XMMRegister dst, Address src, int vector_len) {
6059   assert(VM_Version::supports_evex(), "");
6060   assert(dst != xnoreg, "sanity");
6061   InstructionMark im(this);
6062   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
6063   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_8bit);
6064   // swap src<->dst for encoding
6065   vex_prefix(src, dst->encoding(), dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
6066   emit_int8(0x78);
6067   emit_operand(dst, src);
6068 }
6069 
6070 // duplicate 2-byte integer data from src into 8|16||32 locations in dest : requires AVX512BW and AVX512VL
6071 void Assembler::evpbroadcastw(XMMRegister dst, XMMRegister src, int vector_len) {
6072   assert(VM_Version::supports_evex(), "");
6073   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
6074   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
6075   emit_int8(0x79);
6076   emit_int8((unsigned char)(0xC0 | encode));
6077 }
6078 
6079 void Assembler::evpbroadcastw(XMMRegister dst, Address src, int vector_len) {
6080   assert(VM_Version::supports_evex(), "");
6081   assert(dst != xnoreg, "sanity");
6082   InstructionMark im(this);
6083   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
6084   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_16bit);
6085   // swap src<->dst for encoding
6086   vex_prefix(src, dst->encoding(), dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
6087   emit_int8(0x79);
6088   emit_operand(dst, src);
6089 }
6090 
6091 // duplicate 4-byte integer data from src into 4|8|16 locations in dest : requires AVX512VL
6092 void Assembler::evpbroadcastd(XMMRegister dst, XMMRegister src, int vector_len) {
6093   assert(VM_Version::supports_evex(), "");
6094   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
6095   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
6096   emit_int8(0x58);
6097   emit_int8((unsigned char)(0xC0 | encode));
6098 }
6099 
6100 void Assembler::evpbroadcastd(XMMRegister dst, Address src, int vector_len) {
6101   assert(VM_Version::supports_evex(), "");
6102   assert(dst != xnoreg, "sanity");
6103   InstructionMark im(this);
6104   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
6105   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
6106   // swap src<->dst for encoding
6107   vex_prefix(src, dst->encoding(), dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
6108   emit_int8(0x58);
6109   emit_operand(dst, src);
6110 }
6111 
6112 // duplicate 8-byte integer data from src into 4|8|16 locations in dest : requires AVX512VL
6113 void Assembler::evpbroadcastq(XMMRegister dst, XMMRegister src, int vector_len) {
6114   assert(VM_Version::supports_evex(), "");
6115   InstructionAttr attributes(vector_len, /* vex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
6116   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
6117   emit_int8(0x59);
6118   emit_int8((unsigned char)(0xC0 | encode));
6119 }
6120 
6121 void Assembler::evpbroadcastq(XMMRegister dst, Address src, int vector_len) {
6122   assert(VM_Version::supports_evex(), "");
6123   assert(dst != xnoreg, "sanity");
6124   InstructionMark im(this);
6125   InstructionAttr attributes(vector_len, /* vex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
6126   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
6127   // swap src<->dst for encoding
6128   vex_prefix(src, dst->encoding(), dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
6129   emit_int8(0x59);
6130   emit_operand(dst, src);
6131 }
6132 
6133 // duplicate single precision fp from src into 4|8|16 locations in dest : requires AVX512VL
6134 void Assembler::evpbroadcastss(XMMRegister dst, XMMRegister src, int vector_len) {
6135   assert(VM_Version::supports_evex(), "");
6136   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
6137   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
6138   emit_int8(0x18);
6139   emit_int8((unsigned char)(0xC0 | encode));
6140 }
6141 
6142 void Assembler::evpbroadcastss(XMMRegister dst, Address src, int vector_len) {
6143   assert(VM_Version::supports_evex(), "");
6144   assert(dst != xnoreg, "sanity");
6145   InstructionMark im(this);
6146   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
6147   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_32bit);
6148   // swap src<->dst for encoding
6149   vex_prefix(src, 0, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
6150   emit_int8(0x18);
6151   emit_operand(dst, src);
6152 }
6153 
6154 // duplicate double precision fp from src into 2|4|8 locations in dest : requires AVX512VL
6155 void Assembler::evpbroadcastsd(XMMRegister dst, XMMRegister src, int vector_len) {
6156   assert(VM_Version::supports_evex(), "");
6157   InstructionAttr attributes(vector_len, /* vex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
6158   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
6159   emit_int8(0x19);
6160   emit_int8((unsigned char)(0xC0 | encode));
6161 }
6162 
6163 void Assembler::evpbroadcastsd(XMMRegister dst, Address src, int vector_len) {
6164   assert(VM_Version::supports_evex(), "");
6165   assert(dst != xnoreg, "sanity");
6166   InstructionMark im(this);
6167   InstructionAttr attributes(vector_len, /* vex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
6168   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
6169   // swap src<->dst for encoding
6170   vex_prefix(src, 0, dst->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
6171   emit_int8(0x19);
6172   emit_operand(dst, src);
6173 }
6174 
6175 // duplicate 1-byte integer data from src into 16||32|64 locations in dest : requires AVX512BW and AVX512VL
6176 void Assembler::evpbroadcastb(XMMRegister dst, Register src, int vector_len) {
6177   assert(VM_Version::supports_evex(), "");
6178   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
6179   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
6180   if (attributes.is_evex_instruction()) {
6181     emit_int8(0x7A);
6182   } else {
6183     emit_int8(0x78);
6184   }
6185   emit_int8((unsigned char)(0xC0 | encode));
6186 }
6187 
6188 // duplicate 2-byte integer data from src into 8|16||32 locations in dest : requires AVX512BW and AVX512VL
6189 void Assembler::evpbroadcastw(XMMRegister dst, Register src, int vector_len) {
6190   assert(VM_Version::supports_evex(), "");
6191   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ _legacy_mode_bw, /* no_mask_reg */ true, /* uses_vl */ true);
6192   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
6193   if (attributes.is_evex_instruction()) {
6194     emit_int8(0x7B);
6195   } else {
6196     emit_int8(0x79);
6197   }
6198   emit_int8((unsigned char)(0xC0 | encode));
6199 }
6200 
6201 // duplicate 4-byte integer data from src into 4|8|16 locations in dest : requires AVX512VL
6202 void Assembler::evpbroadcastd(XMMRegister dst, Register src, int vector_len) {
6203   assert(VM_Version::supports_evex(), "");
6204   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
6205   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
6206   if (attributes.is_evex_instruction()) {
6207     emit_int8(0x7C);
6208   } else {
6209     emit_int8(0x58);
6210   }
6211   emit_int8((unsigned char)(0xC0 | encode));
6212 }
6213 
6214 // duplicate 8-byte integer data from src into 4|8|16 locations in dest : requires AVX512VL
6215 void Assembler::evpbroadcastq(XMMRegister dst, Register src, int vector_len) {
6216   assert(VM_Version::supports_evex(), "");
6217   InstructionAttr attributes(vector_len, /* vex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ true);
6218   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
6219   if (attributes.is_evex_instruction()) {
6220     emit_int8(0x7C);
6221   } else {
6222     emit_int8(0x59);
6223   }
6224   emit_int8((unsigned char)(0xC0 | encode));
6225 }
6226 
6227 // Carry-Less Multiplication Quadword
6228 void Assembler::pclmulqdq(XMMRegister dst, XMMRegister src, int mask) {
6229   assert(VM_Version::supports_clmul(), "");
6230   InstructionAttr attributes(AVX_128bit, /* rex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
6231   int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
6232   emit_int8(0x44);
6233   emit_int8((unsigned char)(0xC0 | encode));
6234   emit_int8((unsigned char)mask);
6235 }
6236 
6237 // Carry-Less Multiplication Quadword
6238 void Assembler::vpclmulqdq(XMMRegister dst, XMMRegister nds, XMMRegister src, int mask) {
6239   assert(VM_Version::supports_avx() && VM_Version::supports_clmul(), "");
6240   InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
6241   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
6242   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
6243   emit_int8(0x44);
6244   emit_int8((unsigned char)(0xC0 | encode));
6245   emit_int8((unsigned char)mask);
6246 }
6247 
6248 void Assembler::vzeroupper() {
6249   assert(VM_Version::supports_avx(), "");
6250   InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
6251   (void)vex_prefix_and_encode(0, 0, 0, VEX_SIMD_NONE, VEX_OPCODE_0F, &attributes);
6252   emit_int8(0x77);
6253 }
6254 
6255 
6256 #ifndef _LP64
6257 // 32bit only pieces of the assembler
6258 
6259 void Assembler::cmp_literal32(Register src1, int32_t imm32, RelocationHolder const& rspec) {
6260   // NO PREFIX AS NEVER 64BIT
6261   InstructionMark im(this);
6262   emit_int8((unsigned char)0x81);
6263   emit_int8((unsigned char)(0xF8 | src1->encoding()));
6264   emit_data(imm32, rspec, 0);
6265 }
6266 
6267 void Assembler::cmp_literal32(Address src1, int32_t imm32, RelocationHolder const& rspec) {
6268   // NO PREFIX AS NEVER 64BIT (not even 32bit versions of 64bit regs
6269   InstructionMark im(this);
6270   emit_int8((unsigned char)0x81);
6271   emit_operand(rdi, src1);
6272   emit_data(imm32, rspec, 0);
6273 }
6274 
6275 // The 64-bit (32bit platform) cmpxchg compares the value at adr with the contents of rdx:rax,
6276 // and stores rcx:rbx into adr if so; otherwise, the value at adr is loaded
6277 // into rdx:rax.  The ZF is set if the compared values were equal, and cleared otherwise.
6278 void Assembler::cmpxchg8(Address adr) {
6279   InstructionMark im(this);
6280   emit_int8(0x0F);
6281   emit_int8((unsigned char)0xC7);
6282   emit_operand(rcx, adr);
6283 }
6284 
6285 void Assembler::decl(Register dst) {
6286   // Don't use it directly. Use MacroAssembler::decrementl() instead.
6287  emit_int8(0x48 | dst->encoding());
6288 }
6289 
6290 #endif // _LP64
6291 
6292 // 64bit typically doesn't use the x87 but needs to for the trig funcs
6293 
6294 void Assembler::fabs() {
6295   emit_int8((unsigned char)0xD9);
6296   emit_int8((unsigned char)0xE1);
6297 }
6298 
6299 void Assembler::fadd(int i) {
6300   emit_farith(0xD8, 0xC0, i);
6301 }
6302 
6303 void Assembler::fadd_d(Address src) {
6304   InstructionMark im(this);
6305   emit_int8((unsigned char)0xDC);
6306   emit_operand32(rax, src);
6307 }
6308 
6309 void Assembler::fadd_s(Address src) {
6310   InstructionMark im(this);
6311   emit_int8((unsigned char)0xD8);
6312   emit_operand32(rax, src);
6313 }
6314 
6315 void Assembler::fadda(int i) {
6316   emit_farith(0xDC, 0xC0, i);
6317 }
6318 
6319 void Assembler::faddp(int i) {
6320   emit_farith(0xDE, 0xC0, i);
6321 }
6322 
6323 void Assembler::fchs() {
6324   emit_int8((unsigned char)0xD9);
6325   emit_int8((unsigned char)0xE0);
6326 }
6327 
6328 void Assembler::fcom(int i) {
6329   emit_farith(0xD8, 0xD0, i);
6330 }
6331 
6332 void Assembler::fcomp(int i) {
6333   emit_farith(0xD8, 0xD8, i);
6334 }
6335 
6336 void Assembler::fcomp_d(Address src) {
6337   InstructionMark im(this);
6338   emit_int8((unsigned char)0xDC);
6339   emit_operand32(rbx, src);
6340 }
6341 
6342 void Assembler::fcomp_s(Address src) {
6343   InstructionMark im(this);
6344   emit_int8((unsigned char)0xD8);
6345   emit_operand32(rbx, src);
6346 }
6347 
6348 void Assembler::fcompp() {
6349   emit_int8((unsigned char)0xDE);
6350   emit_int8((unsigned char)0xD9);
6351 }
6352 
6353 void Assembler::fcos() {
6354   emit_int8((unsigned char)0xD9);
6355   emit_int8((unsigned char)0xFF);
6356 }
6357 
6358 void Assembler::fdecstp() {
6359   emit_int8((unsigned char)0xD9);
6360   emit_int8((unsigned char)0xF6);
6361 }
6362 
6363 void Assembler::fdiv(int i) {
6364   emit_farith(0xD8, 0xF0, i);
6365 }
6366 
6367 void Assembler::fdiv_d(Address src) {
6368   InstructionMark im(this);
6369   emit_int8((unsigned char)0xDC);
6370   emit_operand32(rsi, src);
6371 }
6372 
6373 void Assembler::fdiv_s(Address src) {
6374   InstructionMark im(this);
6375   emit_int8((unsigned char)0xD8);
6376   emit_operand32(rsi, src);
6377 }
6378 
6379 void Assembler::fdiva(int i) {
6380   emit_farith(0xDC, 0xF8, i);
6381 }
6382 
6383 // Note: The Intel manual (Pentium Processor User's Manual, Vol.3, 1994)
6384 //       is erroneous for some of the floating-point instructions below.
6385 
6386 void Assembler::fdivp(int i) {
6387   emit_farith(0xDE, 0xF8, i);                    // ST(0) <- ST(0) / ST(1) and pop (Intel manual wrong)
6388 }
6389 
6390 void Assembler::fdivr(int i) {
6391   emit_farith(0xD8, 0xF8, i);
6392 }
6393 
6394 void Assembler::fdivr_d(Address src) {
6395   InstructionMark im(this);
6396   emit_int8((unsigned char)0xDC);
6397   emit_operand32(rdi, src);
6398 }
6399 
6400 void Assembler::fdivr_s(Address src) {
6401   InstructionMark im(this);
6402   emit_int8((unsigned char)0xD8);
6403   emit_operand32(rdi, src);
6404 }
6405 
6406 void Assembler::fdivra(int i) {
6407   emit_farith(0xDC, 0xF0, i);
6408 }
6409 
6410 void Assembler::fdivrp(int i) {
6411   emit_farith(0xDE, 0xF0, i);                    // ST(0) <- ST(1) / ST(0) and pop (Intel manual wrong)
6412 }
6413 
6414 void Assembler::ffree(int i) {
6415   emit_farith(0xDD, 0xC0, i);
6416 }
6417 
6418 void Assembler::fild_d(Address adr) {
6419   InstructionMark im(this);
6420   emit_int8((unsigned char)0xDF);
6421   emit_operand32(rbp, adr);
6422 }
6423 
6424 void Assembler::fild_s(Address adr) {
6425   InstructionMark im(this);
6426   emit_int8((unsigned char)0xDB);
6427   emit_operand32(rax, adr);
6428 }
6429 
6430 void Assembler::fincstp() {
6431   emit_int8((unsigned char)0xD9);
6432   emit_int8((unsigned char)0xF7);
6433 }
6434 
6435 void Assembler::finit() {
6436   emit_int8((unsigned char)0x9B);
6437   emit_int8((unsigned char)0xDB);
6438   emit_int8((unsigned char)0xE3);
6439 }
6440 
6441 void Assembler::fist_s(Address adr) {
6442   InstructionMark im(this);
6443   emit_int8((unsigned char)0xDB);
6444   emit_operand32(rdx, adr);
6445 }
6446 
6447 void Assembler::fistp_d(Address adr) {
6448   InstructionMark im(this);
6449   emit_int8((unsigned char)0xDF);
6450   emit_operand32(rdi, adr);
6451 }
6452 
6453 void Assembler::fistp_s(Address adr) {
6454   InstructionMark im(this);
6455   emit_int8((unsigned char)0xDB);
6456   emit_operand32(rbx, adr);
6457 }
6458 
6459 void Assembler::fld1() {
6460   emit_int8((unsigned char)0xD9);
6461   emit_int8((unsigned char)0xE8);
6462 }
6463 
6464 void Assembler::fld_d(Address adr) {
6465   InstructionMark im(this);
6466   emit_int8((unsigned char)0xDD);
6467   emit_operand32(rax, adr);
6468 }
6469 
6470 void Assembler::fld_s(Address adr) {
6471   InstructionMark im(this);
6472   emit_int8((unsigned char)0xD9);
6473   emit_operand32(rax, adr);
6474 }
6475 
6476 
6477 void Assembler::fld_s(int index) {
6478   emit_farith(0xD9, 0xC0, index);
6479 }
6480 
6481 void Assembler::fld_x(Address adr) {
6482   InstructionMark im(this);
6483   emit_int8((unsigned char)0xDB);
6484   emit_operand32(rbp, adr);
6485 }
6486 
6487 void Assembler::fldcw(Address src) {
6488   InstructionMark im(this);
6489   emit_int8((unsigned char)0xD9);
6490   emit_operand32(rbp, src);
6491 }
6492 
6493 void Assembler::fldenv(Address src) {
6494   InstructionMark im(this);
6495   emit_int8((unsigned char)0xD9);
6496   emit_operand32(rsp, src);
6497 }
6498 
6499 void Assembler::fldlg2() {
6500   emit_int8((unsigned char)0xD9);
6501   emit_int8((unsigned char)0xEC);
6502 }
6503 
6504 void Assembler::fldln2() {
6505   emit_int8((unsigned char)0xD9);
6506   emit_int8((unsigned char)0xED);
6507 }
6508 
6509 void Assembler::fldz() {
6510   emit_int8((unsigned char)0xD9);
6511   emit_int8((unsigned char)0xEE);
6512 }
6513 
6514 void Assembler::flog() {
6515   fldln2();
6516   fxch();
6517   fyl2x();
6518 }
6519 
6520 void Assembler::flog10() {
6521   fldlg2();
6522   fxch();
6523   fyl2x();
6524 }
6525 
6526 void Assembler::fmul(int i) {
6527   emit_farith(0xD8, 0xC8, i);
6528 }
6529 
6530 void Assembler::fmul_d(Address src) {
6531   InstructionMark im(this);
6532   emit_int8((unsigned char)0xDC);
6533   emit_operand32(rcx, src);
6534 }
6535 
6536 void Assembler::fmul_s(Address src) {
6537   InstructionMark im(this);
6538   emit_int8((unsigned char)0xD8);
6539   emit_operand32(rcx, src);
6540 }
6541 
6542 void Assembler::fmula(int i) {
6543   emit_farith(0xDC, 0xC8, i);
6544 }
6545 
6546 void Assembler::fmulp(int i) {
6547   emit_farith(0xDE, 0xC8, i);
6548 }
6549 
6550 void Assembler::fnsave(Address dst) {
6551   InstructionMark im(this);
6552   emit_int8((unsigned char)0xDD);
6553   emit_operand32(rsi, dst);
6554 }
6555 
6556 void Assembler::fnstcw(Address src) {
6557   InstructionMark im(this);
6558   emit_int8((unsigned char)0x9B);
6559   emit_int8((unsigned char)0xD9);
6560   emit_operand32(rdi, src);
6561 }
6562 
6563 void Assembler::fnstsw_ax() {
6564   emit_int8((unsigned char)0xDF);
6565   emit_int8((unsigned char)0xE0);
6566 }
6567 
6568 void Assembler::fprem() {
6569   emit_int8((unsigned char)0xD9);
6570   emit_int8((unsigned char)0xF8);
6571 }
6572 
6573 void Assembler::fprem1() {
6574   emit_int8((unsigned char)0xD9);
6575   emit_int8((unsigned char)0xF5);
6576 }
6577 
6578 void Assembler::frstor(Address src) {
6579   InstructionMark im(this);
6580   emit_int8((unsigned char)0xDD);
6581   emit_operand32(rsp, src);
6582 }
6583 
6584 void Assembler::fsin() {
6585   emit_int8((unsigned char)0xD9);
6586   emit_int8((unsigned char)0xFE);
6587 }
6588 
6589 void Assembler::fsqrt() {
6590   emit_int8((unsigned char)0xD9);
6591   emit_int8((unsigned char)0xFA);
6592 }
6593 
6594 void Assembler::fst_d(Address adr) {
6595   InstructionMark im(this);
6596   emit_int8((unsigned char)0xDD);
6597   emit_operand32(rdx, adr);
6598 }
6599 
6600 void Assembler::fst_s(Address adr) {
6601   InstructionMark im(this);
6602   emit_int8((unsigned char)0xD9);
6603   emit_operand32(rdx, adr);
6604 }
6605 
6606 void Assembler::fstp_d(Address adr) {
6607   InstructionMark im(this);
6608   emit_int8((unsigned char)0xDD);
6609   emit_operand32(rbx, adr);
6610 }
6611 
6612 void Assembler::fstp_d(int index) {
6613   emit_farith(0xDD, 0xD8, index);
6614 }
6615 
6616 void Assembler::fstp_s(Address adr) {
6617   InstructionMark im(this);
6618   emit_int8((unsigned char)0xD9);
6619   emit_operand32(rbx, adr);
6620 }
6621 
6622 void Assembler::fstp_x(Address adr) {
6623   InstructionMark im(this);
6624   emit_int8((unsigned char)0xDB);
6625   emit_operand32(rdi, adr);
6626 }
6627 
6628 void Assembler::fsub(int i) {
6629   emit_farith(0xD8, 0xE0, i);
6630 }
6631 
6632 void Assembler::fsub_d(Address src) {
6633   InstructionMark im(this);
6634   emit_int8((unsigned char)0xDC);
6635   emit_operand32(rsp, src);
6636 }
6637 
6638 void Assembler::fsub_s(Address src) {
6639   InstructionMark im(this);
6640   emit_int8((unsigned char)0xD8);
6641   emit_operand32(rsp, src);
6642 }
6643 
6644 void Assembler::fsuba(int i) {
6645   emit_farith(0xDC, 0xE8, i);
6646 }
6647 
6648 void Assembler::fsubp(int i) {
6649   emit_farith(0xDE, 0xE8, i);                    // ST(0) <- ST(0) - ST(1) and pop (Intel manual wrong)
6650 }
6651 
6652 void Assembler::fsubr(int i) {
6653   emit_farith(0xD8, 0xE8, i);
6654 }
6655 
6656 void Assembler::fsubr_d(Address src) {
6657   InstructionMark im(this);
6658   emit_int8((unsigned char)0xDC);
6659   emit_operand32(rbp, src);
6660 }
6661 
6662 void Assembler::fsubr_s(Address src) {
6663   InstructionMark im(this);
6664   emit_int8((unsigned char)0xD8);
6665   emit_operand32(rbp, src);
6666 }
6667 
6668 void Assembler::fsubra(int i) {
6669   emit_farith(0xDC, 0xE0, i);
6670 }
6671 
6672 void Assembler::fsubrp(int i) {
6673   emit_farith(0xDE, 0xE0, i);                    // ST(0) <- ST(1) - ST(0) and pop (Intel manual wrong)
6674 }
6675 
6676 void Assembler::ftan() {
6677   emit_int8((unsigned char)0xD9);
6678   emit_int8((unsigned char)0xF2);
6679   emit_int8((unsigned char)0xDD);
6680   emit_int8((unsigned char)0xD8);
6681 }
6682 
6683 void Assembler::ftst() {
6684   emit_int8((unsigned char)0xD9);
6685   emit_int8((unsigned char)0xE4);
6686 }
6687 
6688 void Assembler::fucomi(int i) {
6689   // make sure the instruction is supported (introduced for P6, together with cmov)
6690   guarantee(VM_Version::supports_cmov(), "illegal instruction");
6691   emit_farith(0xDB, 0xE8, i);
6692 }
6693 
6694 void Assembler::fucomip(int i) {
6695   // make sure the instruction is supported (introduced for P6, together with cmov)
6696   guarantee(VM_Version::supports_cmov(), "illegal instruction");
6697   emit_farith(0xDF, 0xE8, i);
6698 }
6699 
6700 void Assembler::fwait() {
6701   emit_int8((unsigned char)0x9B);
6702 }
6703 
6704 void Assembler::fxch(int i) {
6705   emit_farith(0xD9, 0xC8, i);
6706 }
6707 
6708 void Assembler::fyl2x() {
6709   emit_int8((unsigned char)0xD9);
6710   emit_int8((unsigned char)0xF1);
6711 }
6712 
6713 void Assembler::frndint() {
6714   emit_int8((unsigned char)0xD9);
6715   emit_int8((unsigned char)0xFC);
6716 }
6717 
6718 void Assembler::f2xm1() {
6719   emit_int8((unsigned char)0xD9);
6720   emit_int8((unsigned char)0xF0);
6721 }
6722 
6723 void Assembler::fldl2e() {
6724   emit_int8((unsigned char)0xD9);
6725   emit_int8((unsigned char)0xEA);
6726 }
6727 
6728 // SSE SIMD prefix byte values corresponding to VexSimdPrefix encoding.
6729 static int simd_pre[4] = { 0, 0x66, 0xF3, 0xF2 };
6730 // SSE opcode second byte values (first is 0x0F) corresponding to VexOpcode encoding.
6731 static int simd_opc[4] = { 0,    0, 0x38, 0x3A };
6732 
6733 // Generate SSE legacy REX prefix and SIMD opcode based on VEX encoding.
6734 void Assembler::rex_prefix(Address adr, XMMRegister xreg, VexSimdPrefix pre, VexOpcode opc, bool rex_w) {
6735   if (pre > 0) {
6736     emit_int8(simd_pre[pre]);
6737   }
6738   if (rex_w) {
6739     prefixq(adr, xreg);
6740   } else {
6741     prefix(adr, xreg);
6742   }
6743   if (opc > 0) {
6744     emit_int8(0x0F);
6745     int opc2 = simd_opc[opc];
6746     if (opc2 > 0) {
6747       emit_int8(opc2);
6748     }
6749   }
6750 }
6751 
6752 int Assembler::rex_prefix_and_encode(int dst_enc, int src_enc, VexSimdPrefix pre, VexOpcode opc, bool rex_w) {
6753   if (pre > 0) {
6754     emit_int8(simd_pre[pre]);
6755   }
6756   int encode = (rex_w) ? prefixq_and_encode(dst_enc, src_enc) : prefix_and_encode(dst_enc, src_enc);
6757   if (opc > 0) {
6758     emit_int8(0x0F);
6759     int opc2 = simd_opc[opc];
6760     if (opc2 > 0) {
6761       emit_int8(opc2);
6762     }
6763   }
6764   return encode;
6765 }
6766 
6767 
6768 void Assembler::vex_prefix(bool vex_r, bool vex_b, bool vex_x, int nds_enc, VexSimdPrefix pre, VexOpcode opc) {
6769   int vector_len = _attributes->get_vector_len();
6770   bool vex_w = _attributes->is_rex_vex_w();
6771   if (vex_b || vex_x || vex_w || (opc == VEX_OPCODE_0F_38) || (opc == VEX_OPCODE_0F_3A)) {
6772     prefix(VEX_3bytes);
6773 
6774     int byte1 = (vex_r ? VEX_R : 0) | (vex_x ? VEX_X : 0) | (vex_b ? VEX_B : 0);
6775     byte1 = (~byte1) & 0xE0;
6776     byte1 |= opc;
6777     emit_int8(byte1);
6778 
6779     int byte2 = ((~nds_enc) & 0xf) << 3;
6780     byte2 |= (vex_w ? VEX_W : 0) | ((vector_len > 0) ? 4 : 0) | pre;
6781     emit_int8(byte2);
6782   } else {
6783     prefix(VEX_2bytes);
6784 
6785     int byte1 = vex_r ? VEX_R : 0;
6786     byte1 = (~byte1) & 0x80;
6787     byte1 |= ((~nds_enc) & 0xf) << 3;
6788     byte1 |= ((vector_len > 0 ) ? 4 : 0) | pre;
6789     emit_int8(byte1);
6790   }
6791 }
6792 
6793 // This is a 4 byte encoding
6794 void Assembler::evex_prefix(bool vex_r, bool vex_b, bool vex_x, bool evex_r, bool evex_v, int nds_enc, VexSimdPrefix pre, VexOpcode opc){
6795   // EVEX 0x62 prefix
6796   prefix(EVEX_4bytes);
6797   bool vex_w = _attributes->is_rex_vex_w();
6798   int evex_encoding = (vex_w ? VEX_W : 0);
6799   // EVEX.b is not currently used for broadcast of single element or data rounding modes
6800   _attributes->set_evex_encoding(evex_encoding);
6801 
6802   // P0: byte 2, initialized to RXBR`00mm
6803   // instead of not'd
6804   int byte2 = (vex_r ? VEX_R : 0) | (vex_x ? VEX_X : 0) | (vex_b ? VEX_B : 0) | (evex_r ? EVEX_Rb : 0);
6805   byte2 = (~byte2) & 0xF0;
6806   // confine opc opcode extensions in mm bits to lower two bits
6807   // of form {0F, 0F_38, 0F_3A}
6808   byte2 |= opc;
6809   emit_int8(byte2);
6810 
6811   // P1: byte 3 as Wvvvv1pp
6812   int byte3 = ((~nds_enc) & 0xf) << 3;
6813   // p[10] is always 1
6814   byte3 |= EVEX_F;
6815   byte3 |= (vex_w & 1) << 7;
6816   // confine pre opcode extensions in pp bits to lower two bits
6817   // of form {66, F3, F2}
6818   byte3 |= pre;
6819   emit_int8(byte3);
6820 
6821   // P2: byte 4 as zL'Lbv'aaa
6822   int byte4 = (_attributes->is_no_reg_mask()) ? 0 : 1; // kregs are implemented in the low 3 bits as aaa (hard code k1, it will be initialized for now)
6823   // EVEX.v` for extending EVEX.vvvv or VIDX
6824   byte4 |= (evex_v ? 0: EVEX_V);
6825   // third EXEC.b for broadcast actions
6826   byte4 |= (_attributes->is_extended_context() ? EVEX_Rb : 0);
6827   // fourth EVEX.L'L for vector length : 0 is 128, 1 is 256, 2 is 512, currently we do not support 1024
6828   byte4 |= ((_attributes->get_vector_len())& 0x3) << 5;
6829   // last is EVEX.z for zero/merge actions
6830   byte4 |= (_attributes->is_clear_context() ? EVEX_Z : 0);
6831   emit_int8(byte4);
6832 }
6833 
6834 void Assembler::vex_prefix(Address adr, int nds_enc, int xreg_enc, VexSimdPrefix pre, VexOpcode opc, InstructionAttr *attributes) {
6835   bool vex_r = ((xreg_enc & 8) == 8) ? 1 : 0;
6836   bool vex_b = adr.base_needs_rex();
6837   bool vex_x = adr.index_needs_rex();
6838   set_attributes(attributes);
6839   attributes->set_current_assembler(this);
6840 
6841   // if vector length is turned off, revert to AVX for vectors smaller than 512-bit
6842   if (UseAVX > 2 && _legacy_mode_vl && attributes->uses_vl()) {
6843     switch (attributes->get_vector_len()) {
6844     case AVX_128bit:
6845     case AVX_256bit:
6846       attributes->set_is_legacy_mode();
6847       break;
6848     }
6849   }
6850 
6851   // For pure EVEX check and see if this instruction
6852   // is allowed in legacy mode and has resources which will
6853   // fit in it.  Pure EVEX instructions will use set_is_evex_instruction in their definition,
6854   // else that field is set when we encode to EVEX
6855   if (UseAVX > 2 && !attributes->is_legacy_mode() &&
6856       !_is_managed && !attributes->is_evex_instruction()) {
6857     if (!_legacy_mode_vl && attributes->get_vector_len() != AVX_512bit) {
6858       bool check_register_bank = NOT_IA32(true) IA32_ONLY(false);
6859       if (check_register_bank) {
6860         // check nds_enc and xreg_enc for upper bank usage
6861         if (nds_enc < 16 && xreg_enc < 16) {
6862           attributes->set_is_legacy_mode();
6863         }
6864       } else {
6865         attributes->set_is_legacy_mode();
6866       }
6867     }
6868   }
6869 
6870   _is_managed = false;
6871   if (UseAVX > 2 && !attributes->is_legacy_mode())
6872   {
6873     bool evex_r = (xreg_enc >= 16);
6874     bool evex_v = (nds_enc >= 16);
6875     attributes->set_is_evex_instruction();
6876     evex_prefix(vex_r, vex_b, vex_x, evex_r, evex_v, nds_enc, pre, opc);
6877   } else {
6878     vex_prefix(vex_r, vex_b, vex_x, nds_enc, pre, opc);
6879   }
6880 }
6881 
6882 int Assembler::vex_prefix_and_encode(int dst_enc, int nds_enc, int src_enc, VexSimdPrefix pre, VexOpcode opc, InstructionAttr *attributes) {
6883   bool vex_r = ((dst_enc & 8) == 8) ? 1 : 0;
6884   bool vex_b = ((src_enc & 8) == 8) ? 1 : 0;
6885   bool vex_x = false;
6886   set_attributes(attributes);
6887   attributes->set_current_assembler(this);
6888   bool check_register_bank = NOT_IA32(true) IA32_ONLY(false);
6889 
6890   // if vector length is turned off, revert to AVX for vectors smaller than 512-bit
6891   if (UseAVX > 2 && _legacy_mode_vl && attributes->uses_vl()) {
6892     switch (attributes->get_vector_len()) {
6893     case AVX_128bit:
6894     case AVX_256bit:
6895       if (check_register_bank) {
6896         if (dst_enc >= 16 || nds_enc >= 16 || src_enc >= 16) {
6897           // up propagate arithmetic instructions to meet RA requirements
6898           attributes->set_vector_len(AVX_512bit);
6899         } else {
6900           attributes->set_is_legacy_mode();
6901         }
6902       } else {
6903         attributes->set_is_legacy_mode();
6904       }
6905       break;
6906     }
6907   }
6908 
6909   // For pure EVEX check and see if this instruction
6910   // is allowed in legacy mode and has resources which will
6911   // fit in it.  Pure EVEX instructions will use set_is_evex_instruction in their definition,
6912   // else that field is set when we encode to EVEX
6913   if (UseAVX > 2 && !attributes->is_legacy_mode() &&
6914       !_is_managed && !attributes->is_evex_instruction()) {
6915     if (!_legacy_mode_vl && attributes->get_vector_len() != AVX_512bit) {
6916       if (check_register_bank) {
6917         // check dst_enc, nds_enc and src_enc for upper bank usage
6918         if (dst_enc < 16 && nds_enc < 16 && src_enc < 16) {
6919           attributes->set_is_legacy_mode();
6920         }
6921       } else {
6922         attributes->set_is_legacy_mode();
6923       }
6924     }
6925   }
6926 
6927   _is_managed = false;
6928   if (UseAVX > 2 && !attributes->is_legacy_mode())
6929   {
6930     bool evex_r = (dst_enc >= 16);
6931     bool evex_v = (nds_enc >= 16);
6932     // can use vex_x as bank extender on rm encoding
6933     vex_x = (src_enc >= 16);
6934     attributes->set_is_evex_instruction();
6935     evex_prefix(vex_r, vex_b, vex_x, evex_r, evex_v, nds_enc, pre, opc);
6936   } else {
6937     vex_prefix(vex_r, vex_b, vex_x, nds_enc, pre, opc);
6938   }
6939 
6940   // return modrm byte components for operands
6941   return (((dst_enc & 7) << 3) | (src_enc & 7));
6942 }
6943 
6944 
6945 void Assembler::simd_prefix(XMMRegister xreg, XMMRegister nds, Address adr, VexSimdPrefix pre,
6946                             VexOpcode opc, InstructionAttr *attributes) {
6947   if (UseAVX > 0) {
6948     int xreg_enc = xreg->encoding();
6949     int nds_enc = nds->is_valid() ? nds->encoding() : 0;
6950     vex_prefix(adr, nds_enc, xreg_enc, pre, opc, attributes);
6951   } else {
6952     assert((nds == xreg) || (nds == xnoreg), "wrong sse encoding");
6953     rex_prefix(adr, xreg, pre, opc, attributes->is_rex_vex_w());
6954   }
6955 }
6956 
6957 int Assembler::simd_prefix_and_encode(XMMRegister dst, XMMRegister nds, XMMRegister src, VexSimdPrefix pre,
6958                                       VexOpcode opc, InstructionAttr *attributes) {
6959   int dst_enc = dst->encoding();
6960   int src_enc = src->encoding();
6961   if (UseAVX > 0) {
6962     int nds_enc = nds->is_valid() ? nds->encoding() : 0;
6963     return vex_prefix_and_encode(dst_enc, nds_enc, src_enc, pre, opc, attributes);
6964   } else {
6965     assert((nds == dst) || (nds == src) || (nds == xnoreg), "wrong sse encoding");
6966     return rex_prefix_and_encode(dst_enc, src_enc, pre, opc, attributes->is_rex_vex_w());
6967   }
6968 }
6969 
6970 void Assembler::cmppd(XMMRegister dst, XMMRegister nds, XMMRegister src, int cop, int vector_len) {
6971   assert(VM_Version::supports_avx(), "");
6972   assert(!VM_Version::supports_evex(), "");
6973   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
6974   int encode = simd_prefix_and_encode(dst, nds, src, VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
6975   emit_int8((unsigned char)0xC2);
6976   emit_int8((unsigned char)(0xC0 | encode));
6977   emit_int8((unsigned char)(0xF & cop));
6978 }
6979 
6980 void Assembler::vpblendd(XMMRegister dst, XMMRegister nds, XMMRegister src1, XMMRegister src2, int vector_len) {
6981   assert(VM_Version::supports_avx(), "");
6982   assert(!VM_Version::supports_evex(), "");
6983   InstructionAttr attributes(vector_len, /* vex_w */ false, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
6984   int nds_enc = nds->is_valid() ? nds->encoding() : 0;
6985   int encode = vex_prefix_and_encode(dst->encoding(), nds_enc, src1->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_3A, &attributes);
6986   emit_int8((unsigned char)0x4B);
6987   emit_int8((unsigned char)(0xC0 | encode));
6988   int src2_enc = src2->encoding();
6989   emit_int8((unsigned char)(0xF0 & src2_enc<<4));
6990 }
6991 
6992 
6993 #ifndef _LP64
6994 
6995 void Assembler::incl(Register dst) {
6996   // Don't use it directly. Use MacroAssembler::incrementl() instead.
6997   emit_int8(0x40 | dst->encoding());
6998 }
6999 
7000 void Assembler::lea(Register dst, Address src) {
7001   leal(dst, src);
7002 }
7003 
7004 void Assembler::mov_literal32(Address dst, int32_t imm32, RelocationHolder const& rspec) {
7005   InstructionMark im(this);
7006   emit_int8((unsigned char)0xC7);
7007   emit_operand(rax, dst);
7008   emit_data((int)imm32, rspec, 0);
7009 }
7010 
7011 void Assembler::mov_literal32(Register dst, int32_t imm32, RelocationHolder const& rspec) {
7012   InstructionMark im(this);
7013   int encode = prefix_and_encode(dst->encoding());
7014   emit_int8((unsigned char)(0xB8 | encode));
7015   emit_data((int)imm32, rspec, 0);
7016 }
7017 
7018 void Assembler::popa() { // 32bit
7019   emit_int8(0x61);
7020 }
7021 
7022 void Assembler::push_literal32(int32_t imm32, RelocationHolder const& rspec) {
7023   InstructionMark im(this);
7024   emit_int8(0x68);
7025   emit_data(imm32, rspec, 0);
7026 }
7027 
7028 void Assembler::pusha() { // 32bit
7029   emit_int8(0x60);
7030 }
7031 
7032 void Assembler::set_byte_if_not_zero(Register dst) {
7033   emit_int8(0x0F);
7034   emit_int8((unsigned char)0x95);
7035   emit_int8((unsigned char)(0xE0 | dst->encoding()));
7036 }
7037 
7038 void Assembler::shldl(Register dst, Register src) {
7039   emit_int8(0x0F);
7040   emit_int8((unsigned char)0xA5);
7041   emit_int8((unsigned char)(0xC0 | src->encoding() << 3 | dst->encoding()));
7042 }
7043 
7044 // 0F A4 / r ib
7045 void Assembler::shldl(Register dst, Register src, int8_t imm8) {
7046   emit_int8(0x0F);
7047   emit_int8((unsigned char)0xA4);
7048   emit_int8((unsigned char)(0xC0 | src->encoding() << 3 | dst->encoding()));
7049   emit_int8(imm8);
7050 }
7051 
7052 void Assembler::shrdl(Register dst, Register src) {
7053   emit_int8(0x0F);
7054   emit_int8((unsigned char)0xAD);
7055   emit_int8((unsigned char)(0xC0 | src->encoding() << 3 | dst->encoding()));
7056 }
7057 
7058 #else // LP64
7059 
7060 void Assembler::set_byte_if_not_zero(Register dst) {
7061   int enc = prefix_and_encode(dst->encoding(), true);
7062   emit_int8(0x0F);
7063   emit_int8((unsigned char)0x95);
7064   emit_int8((unsigned char)(0xE0 | enc));
7065 }
7066 
7067 // 64bit only pieces of the assembler
7068 // This should only be used by 64bit instructions that can use rip-relative
7069 // it cannot be used by instructions that want an immediate value.
7070 
7071 bool Assembler::reachable(AddressLiteral adr) {
7072   int64_t disp;
7073   // None will force a 64bit literal to the code stream. Likely a placeholder
7074   // for something that will be patched later and we need to certain it will
7075   // always be reachable.
7076   if (adr.reloc() == relocInfo::none) {
7077     return false;
7078   }
7079   if (adr.reloc() == relocInfo::internal_word_type) {
7080     // This should be rip relative and easily reachable.
7081     return true;
7082   }
7083   if (adr.reloc() == relocInfo::virtual_call_type ||
7084       adr.reloc() == relocInfo::opt_virtual_call_type ||
7085       adr.reloc() == relocInfo::static_call_type ||
7086       adr.reloc() == relocInfo::static_stub_type ) {
7087     // This should be rip relative within the code cache and easily
7088     // reachable until we get huge code caches. (At which point
7089     // ic code is going to have issues).
7090     return true;
7091   }
7092   if (adr.reloc() != relocInfo::external_word_type &&
7093       adr.reloc() != relocInfo::poll_return_type &&  // these are really external_word but need special
7094       adr.reloc() != relocInfo::poll_type &&         // relocs to identify them
7095       adr.reloc() != relocInfo::runtime_call_type ) {
7096     return false;
7097   }
7098 
7099   // Stress the correction code
7100   if (ForceUnreachable) {
7101     // Must be runtimecall reloc, see if it is in the codecache
7102     // Flipping stuff in the codecache to be unreachable causes issues
7103     // with things like inline caches where the additional instructions
7104     // are not handled.
7105     if (CodeCache::find_blob(adr._target) == NULL) {
7106       return false;
7107     }
7108   }
7109   // For external_word_type/runtime_call_type if it is reachable from where we
7110   // are now (possibly a temp buffer) and where we might end up
7111   // anywhere in the codeCache then we are always reachable.
7112   // This would have to change if we ever save/restore shared code
7113   // to be more pessimistic.
7114   disp = (int64_t)adr._target - ((int64_t)CodeCache::low_bound() + sizeof(int));
7115   if (!is_simm32(disp)) return false;
7116   disp = (int64_t)adr._target - ((int64_t)CodeCache::high_bound() + sizeof(int));
7117   if (!is_simm32(disp)) return false;
7118 
7119   disp = (int64_t)adr._target - ((int64_t)pc() + sizeof(int));
7120 
7121   // Because rip relative is a disp + address_of_next_instruction and we
7122   // don't know the value of address_of_next_instruction we apply a fudge factor
7123   // to make sure we will be ok no matter the size of the instruction we get placed into.
7124   // We don't have to fudge the checks above here because they are already worst case.
7125 
7126   // 12 == override/rex byte, opcode byte, rm byte, sib byte, a 4-byte disp , 4-byte literal
7127   // + 4 because better safe than sorry.
7128   const int fudge = 12 + 4;
7129   if (disp < 0) {
7130     disp -= fudge;
7131   } else {
7132     disp += fudge;
7133   }
7134   return is_simm32(disp);
7135 }
7136 
7137 // Check if the polling page is not reachable from the code cache using rip-relative
7138 // addressing.
7139 bool Assembler::is_polling_page_far() {
7140   intptr_t addr = (intptr_t)os::get_polling_page();
7141   return ForceUnreachable ||
7142          !is_simm32(addr - (intptr_t)CodeCache::low_bound()) ||
7143          !is_simm32(addr - (intptr_t)CodeCache::high_bound());
7144 }
7145 
7146 void Assembler::emit_data64(jlong data,
7147                             relocInfo::relocType rtype,
7148                             int format) {
7149   if (rtype == relocInfo::none) {
7150     emit_int64(data);
7151   } else {
7152     emit_data64(data, Relocation::spec_simple(rtype), format);
7153   }
7154 }
7155 
7156 void Assembler::emit_data64(jlong data,
7157                             RelocationHolder const& rspec,
7158                             int format) {
7159   assert(imm_operand == 0, "default format must be immediate in this file");
7160   assert(imm_operand == format, "must be immediate");
7161   assert(inst_mark() != NULL, "must be inside InstructionMark");
7162   // Do not use AbstractAssembler::relocate, which is not intended for
7163   // embedded words.  Instead, relocate to the enclosing instruction.
7164   code_section()->relocate(inst_mark(), rspec, format);
7165 #ifdef ASSERT
7166   check_relocation(rspec, format);
7167 #endif
7168   emit_int64(data);
7169 }
7170 
7171 int Assembler::prefix_and_encode(int reg_enc, bool byteinst) {
7172   if (reg_enc >= 8) {
7173     prefix(REX_B);
7174     reg_enc -= 8;
7175   } else if (byteinst && reg_enc >= 4) {
7176     prefix(REX);
7177   }
7178   return reg_enc;
7179 }
7180 
7181 int Assembler::prefixq_and_encode(int reg_enc) {
7182   if (reg_enc < 8) {
7183     prefix(REX_W);
7184   } else {
7185     prefix(REX_WB);
7186     reg_enc -= 8;
7187   }
7188   return reg_enc;
7189 }
7190 
7191 int Assembler::prefix_and_encode(int dst_enc, bool dst_is_byte, int src_enc, bool src_is_byte) {
7192   if (dst_enc < 8) {
7193     if (src_enc >= 8) {
7194       prefix(REX_B);
7195       src_enc -= 8;
7196     } else if ((src_is_byte && src_enc >= 4) || (dst_is_byte && dst_enc >= 4)) {
7197       prefix(REX);
7198     }
7199   } else {
7200     if (src_enc < 8) {
7201       prefix(REX_R);
7202     } else {
7203       prefix(REX_RB);
7204       src_enc -= 8;
7205     }
7206     dst_enc -= 8;
7207   }
7208   return dst_enc << 3 | src_enc;
7209 }
7210 
7211 int Assembler::prefixq_and_encode(int dst_enc, int src_enc) {
7212   if (dst_enc < 8) {
7213     if (src_enc < 8) {
7214       prefix(REX_W);
7215     } else {
7216       prefix(REX_WB);
7217       src_enc -= 8;
7218     }
7219   } else {
7220     if (src_enc < 8) {
7221       prefix(REX_WR);
7222     } else {
7223       prefix(REX_WRB);
7224       src_enc -= 8;
7225     }
7226     dst_enc -= 8;
7227   }
7228   return dst_enc << 3 | src_enc;
7229 }
7230 
7231 void Assembler::prefix(Register reg) {
7232   if (reg->encoding() >= 8) {
7233     prefix(REX_B);
7234   }
7235 }
7236 
7237 void Assembler::prefix(Register dst, Register src, Prefix p) {
7238   if (src->encoding() >= 8) {
7239     p = (Prefix)(p | REX_B);
7240   }
7241   if (dst->encoding() >= 8) {
7242     p = (Prefix)( p | REX_R);
7243   }
7244   if (p != Prefix_EMPTY) {
7245     // do not generate an empty prefix
7246     prefix(p);
7247   }
7248 }
7249 
7250 void Assembler::prefix(Register dst, Address adr, Prefix p) {
7251   if (adr.base_needs_rex()) {
7252     if (adr.index_needs_rex()) {
7253       assert(false, "prefix(Register dst, Address adr, Prefix p) does not support handling of an X");
7254     } else {
7255       prefix(REX_B);
7256     }
7257   } else {
7258     if (adr.index_needs_rex()) {
7259       assert(false, "prefix(Register dst, Address adr, Prefix p) does not support handling of an X");
7260     }
7261   }
7262   if (dst->encoding() >= 8) {
7263     p = (Prefix)(p | REX_R);
7264   }
7265   if (p != Prefix_EMPTY) {
7266     // do not generate an empty prefix
7267     prefix(p);
7268   }
7269 }
7270 
7271 void Assembler::prefix(Address adr) {
7272   if (adr.base_needs_rex()) {
7273     if (adr.index_needs_rex()) {
7274       prefix(REX_XB);
7275     } else {
7276       prefix(REX_B);
7277     }
7278   } else {
7279     if (adr.index_needs_rex()) {
7280       prefix(REX_X);
7281     }
7282   }
7283 }
7284 
7285 void Assembler::prefixq(Address adr) {
7286   if (adr.base_needs_rex()) {
7287     if (adr.index_needs_rex()) {
7288       prefix(REX_WXB);
7289     } else {
7290       prefix(REX_WB);
7291     }
7292   } else {
7293     if (adr.index_needs_rex()) {
7294       prefix(REX_WX);
7295     } else {
7296       prefix(REX_W);
7297     }
7298   }
7299 }
7300 
7301 
7302 void Assembler::prefix(Address adr, Register reg, bool byteinst) {
7303   if (reg->encoding() < 8) {
7304     if (adr.base_needs_rex()) {
7305       if (adr.index_needs_rex()) {
7306         prefix(REX_XB);
7307       } else {
7308         prefix(REX_B);
7309       }
7310     } else {
7311       if (adr.index_needs_rex()) {
7312         prefix(REX_X);
7313       } else if (byteinst && reg->encoding() >= 4 ) {
7314         prefix(REX);
7315       }
7316     }
7317   } else {
7318     if (adr.base_needs_rex()) {
7319       if (adr.index_needs_rex()) {
7320         prefix(REX_RXB);
7321       } else {
7322         prefix(REX_RB);
7323       }
7324     } else {
7325       if (adr.index_needs_rex()) {
7326         prefix(REX_RX);
7327       } else {
7328         prefix(REX_R);
7329       }
7330     }
7331   }
7332 }
7333 
7334 void Assembler::prefixq(Address adr, Register src) {
7335   if (src->encoding() < 8) {
7336     if (adr.base_needs_rex()) {
7337       if (adr.index_needs_rex()) {
7338         prefix(REX_WXB);
7339       } else {
7340         prefix(REX_WB);
7341       }
7342     } else {
7343       if (adr.index_needs_rex()) {
7344         prefix(REX_WX);
7345       } else {
7346         prefix(REX_W);
7347       }
7348     }
7349   } else {
7350     if (adr.base_needs_rex()) {
7351       if (adr.index_needs_rex()) {
7352         prefix(REX_WRXB);
7353       } else {
7354         prefix(REX_WRB);
7355       }
7356     } else {
7357       if (adr.index_needs_rex()) {
7358         prefix(REX_WRX);
7359       } else {
7360         prefix(REX_WR);
7361       }
7362     }
7363   }
7364 }
7365 
7366 void Assembler::prefix(Address adr, XMMRegister reg) {
7367   if (reg->encoding() < 8) {
7368     if (adr.base_needs_rex()) {
7369       if (adr.index_needs_rex()) {
7370         prefix(REX_XB);
7371       } else {
7372         prefix(REX_B);
7373       }
7374     } else {
7375       if (adr.index_needs_rex()) {
7376         prefix(REX_X);
7377       }
7378     }
7379   } else {
7380     if (adr.base_needs_rex()) {
7381       if (adr.index_needs_rex()) {
7382         prefix(REX_RXB);
7383       } else {
7384         prefix(REX_RB);
7385       }
7386     } else {
7387       if (adr.index_needs_rex()) {
7388         prefix(REX_RX);
7389       } else {
7390         prefix(REX_R);
7391       }
7392     }
7393   }
7394 }
7395 
7396 void Assembler::prefixq(Address adr, XMMRegister src) {
7397   if (src->encoding() < 8) {
7398     if (adr.base_needs_rex()) {
7399       if (adr.index_needs_rex()) {
7400         prefix(REX_WXB);
7401       } else {
7402         prefix(REX_WB);
7403       }
7404     } else {
7405       if (adr.index_needs_rex()) {
7406         prefix(REX_WX);
7407       } else {
7408         prefix(REX_W);
7409       }
7410     }
7411   } else {
7412     if (adr.base_needs_rex()) {
7413       if (adr.index_needs_rex()) {
7414         prefix(REX_WRXB);
7415       } else {
7416         prefix(REX_WRB);
7417       }
7418     } else {
7419       if (adr.index_needs_rex()) {
7420         prefix(REX_WRX);
7421       } else {
7422         prefix(REX_WR);
7423       }
7424     }
7425   }
7426 }
7427 
7428 void Assembler::adcq(Register dst, int32_t imm32) {
7429   (void) prefixq_and_encode(dst->encoding());
7430   emit_arith(0x81, 0xD0, dst, imm32);
7431 }
7432 
7433 void Assembler::adcq(Register dst, Address src) {
7434   InstructionMark im(this);
7435   prefixq(src, dst);
7436   emit_int8(0x13);
7437   emit_operand(dst, src);
7438 }
7439 
7440 void Assembler::adcq(Register dst, Register src) {
7441   (void) prefixq_and_encode(dst->encoding(), src->encoding());
7442   emit_arith(0x13, 0xC0, dst, src);
7443 }
7444 
7445 void Assembler::addq(Address dst, int32_t imm32) {
7446   InstructionMark im(this);
7447   prefixq(dst);
7448   emit_arith_operand(0x81, rax, dst,imm32);
7449 }
7450 
7451 void Assembler::addq(Address dst, Register src) {
7452   InstructionMark im(this);
7453   prefixq(dst, src);
7454   emit_int8(0x01);
7455   emit_operand(src, dst);
7456 }
7457 
7458 void Assembler::addq(Register dst, int32_t imm32) {
7459   (void) prefixq_and_encode(dst->encoding());
7460   emit_arith(0x81, 0xC0, dst, imm32);
7461 }
7462 
7463 void Assembler::addq(Register dst, Address src) {
7464   InstructionMark im(this);
7465   prefixq(src, dst);
7466   emit_int8(0x03);
7467   emit_operand(dst, src);
7468 }
7469 
7470 void Assembler::addq(Register dst, Register src) {
7471   (void) prefixq_and_encode(dst->encoding(), src->encoding());
7472   emit_arith(0x03, 0xC0, dst, src);
7473 }
7474 
7475 void Assembler::adcxq(Register dst, Register src) {
7476   //assert(VM_Version::supports_adx(), "adx instructions not supported");
7477   emit_int8((unsigned char)0x66);
7478   int encode = prefixq_and_encode(dst->encoding(), src->encoding());
7479   emit_int8(0x0F);
7480   emit_int8(0x38);
7481   emit_int8((unsigned char)0xF6);
7482   emit_int8((unsigned char)(0xC0 | encode));
7483 }
7484 
7485 void Assembler::adoxq(Register dst, Register src) {
7486   //assert(VM_Version::supports_adx(), "adx instructions not supported");
7487   emit_int8((unsigned char)0xF3);
7488   int encode = prefixq_and_encode(dst->encoding(), src->encoding());
7489   emit_int8(0x0F);
7490   emit_int8(0x38);
7491   emit_int8((unsigned char)0xF6);
7492   emit_int8((unsigned char)(0xC0 | encode));
7493 }
7494 
7495 void Assembler::andq(Address dst, int32_t imm32) {
7496   InstructionMark im(this);
7497   prefixq(dst);
7498   emit_int8((unsigned char)0x81);
7499   emit_operand(rsp, dst, 4);
7500   emit_int32(imm32);
7501 }
7502 
7503 void Assembler::andq(Register dst, int32_t imm32) {
7504   (void) prefixq_and_encode(dst->encoding());
7505   emit_arith(0x81, 0xE0, dst, imm32);
7506 }
7507 
7508 void Assembler::andq(Register dst, Address src) {
7509   InstructionMark im(this);
7510   prefixq(src, dst);
7511   emit_int8(0x23);
7512   emit_operand(dst, src);
7513 }
7514 
7515 void Assembler::andq(Register dst, Register src) {
7516   (void) prefixq_and_encode(dst->encoding(), src->encoding());
7517   emit_arith(0x23, 0xC0, dst, src);
7518 }
7519 
7520 void Assembler::andnq(Register dst, Register src1, Register src2) {
7521   assert(VM_Version::supports_bmi1(), "bit manipulation instructions not supported");
7522   InstructionAttr attributes(AVX_128bit, /* vex_w */ true, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
7523   int encode = vex_prefix_and_encode(dst->encoding(), src1->encoding(), src2->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
7524   emit_int8((unsigned char)0xF2);
7525   emit_int8((unsigned char)(0xC0 | encode));
7526 }
7527 
7528 void Assembler::andnq(Register dst, Register src1, Address src2) {
7529   assert(VM_Version::supports_bmi1(), "bit manipulation instructions not supported");
7530   InstructionMark im(this);
7531   InstructionAttr attributes(AVX_128bit, /* vex_w */ true, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
7532   vex_prefix(src2, src1->encoding(), dst->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
7533   emit_int8((unsigned char)0xF2);
7534   emit_operand(dst, src2);
7535 }
7536 
7537 void Assembler::bsfq(Register dst, Register src) {
7538   int encode = prefixq_and_encode(dst->encoding(), src->encoding());
7539   emit_int8(0x0F);
7540   emit_int8((unsigned char)0xBC);
7541   emit_int8((unsigned char)(0xC0 | encode));
7542 }
7543 
7544 void Assembler::bsrq(Register dst, Register src) {
7545   int encode = prefixq_and_encode(dst->encoding(), src->encoding());
7546   emit_int8(0x0F);
7547   emit_int8((unsigned char)0xBD);
7548   emit_int8((unsigned char)(0xC0 | encode));
7549 }
7550 
7551 void Assembler::bswapq(Register reg) {
7552   int encode = prefixq_and_encode(reg->encoding());
7553   emit_int8(0x0F);
7554   emit_int8((unsigned char)(0xC8 | encode));
7555 }
7556 
7557 void Assembler::blsiq(Register dst, Register src) {
7558   assert(VM_Version::supports_bmi1(), "bit manipulation instructions not supported");
7559   InstructionAttr attributes(AVX_128bit, /* vex_w */ true, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
7560   int encode = vex_prefix_and_encode(rbx->encoding(), dst->encoding(), src->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
7561   emit_int8((unsigned char)0xF3);
7562   emit_int8((unsigned char)(0xC0 | encode));
7563 }
7564 
7565 void Assembler::blsiq(Register dst, Address src) {
7566   assert(VM_Version::supports_bmi1(), "bit manipulation instructions not supported");
7567   InstructionMark im(this);
7568   InstructionAttr attributes(AVX_128bit, /* vex_w */ true, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
7569   vex_prefix(src, dst->encoding(), rbx->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
7570   emit_int8((unsigned char)0xF3);
7571   emit_operand(rbx, src);
7572 }
7573 
7574 void Assembler::blsmskq(Register dst, Register src) {
7575   assert(VM_Version::supports_bmi1(), "bit manipulation instructions not supported");
7576   InstructionAttr attributes(AVX_128bit, /* vex_w */ true, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
7577   int encode = vex_prefix_and_encode(rdx->encoding(), dst->encoding(), src->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
7578   emit_int8((unsigned char)0xF3);
7579   emit_int8((unsigned char)(0xC0 | encode));
7580 }
7581 
7582 void Assembler::blsmskq(Register dst, Address src) {
7583   assert(VM_Version::supports_bmi1(), "bit manipulation instructions not supported");
7584   InstructionMark im(this);
7585   InstructionAttr attributes(AVX_128bit, /* vex_w */ true, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
7586   vex_prefix(src, dst->encoding(), rdx->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
7587   emit_int8((unsigned char)0xF3);
7588   emit_operand(rdx, src);
7589 }
7590 
7591 void Assembler::blsrq(Register dst, Register src) {
7592   assert(VM_Version::supports_bmi1(), "bit manipulation instructions not supported");
7593   InstructionAttr attributes(AVX_128bit, /* vex_w */ true, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
7594   int encode = vex_prefix_and_encode(rcx->encoding(), dst->encoding(), src->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
7595   emit_int8((unsigned char)0xF3);
7596   emit_int8((unsigned char)(0xC0 | encode));
7597 }
7598 
7599 void Assembler::blsrq(Register dst, Address src) {
7600   assert(VM_Version::supports_bmi1(), "bit manipulation instructions not supported");
7601   InstructionMark im(this);
7602   InstructionAttr attributes(AVX_128bit, /* vex_w */ true, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
7603   vex_prefix(src, dst->encoding(), rcx->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F_38, &attributes);
7604   emit_int8((unsigned char)0xF3);
7605   emit_operand(rcx, src);
7606 }
7607 
7608 void Assembler::cdqq() {
7609   prefix(REX_W);
7610   emit_int8((unsigned char)0x99);
7611 }
7612 
7613 void Assembler::clflush(Address adr) {
7614   prefix(adr);
7615   emit_int8(0x0F);
7616   emit_int8((unsigned char)0xAE);
7617   emit_operand(rdi, adr);
7618 }
7619 
7620 void Assembler::cmovq(Condition cc, Register dst, Register src) {
7621   int encode = prefixq_and_encode(dst->encoding(), src->encoding());
7622   emit_int8(0x0F);
7623   emit_int8(0x40 | cc);
7624   emit_int8((unsigned char)(0xC0 | encode));
7625 }
7626 
7627 void Assembler::cmovq(Condition cc, Register dst, Address src) {
7628   InstructionMark im(this);
7629   prefixq(src, dst);
7630   emit_int8(0x0F);
7631   emit_int8(0x40 | cc);
7632   emit_operand(dst, src);
7633 }
7634 
7635 void Assembler::cmpq(Address dst, int32_t imm32) {
7636   InstructionMark im(this);
7637   prefixq(dst);
7638   emit_int8((unsigned char)0x81);
7639   emit_operand(rdi, dst, 4);
7640   emit_int32(imm32);
7641 }
7642 
7643 void Assembler::cmpq(Register dst, int32_t imm32) {
7644   (void) prefixq_and_encode(dst->encoding());
7645   emit_arith(0x81, 0xF8, dst, imm32);
7646 }
7647 
7648 void Assembler::cmpq(Address dst, Register src) {
7649   InstructionMark im(this);
7650   prefixq(dst, src);
7651   emit_int8(0x3B);
7652   emit_operand(src, dst);
7653 }
7654 
7655 void Assembler::cmpq(Register dst, Register src) {
7656   (void) prefixq_and_encode(dst->encoding(), src->encoding());
7657   emit_arith(0x3B, 0xC0, dst, src);
7658 }
7659 
7660 void Assembler::cmpq(Register dst, Address  src) {
7661   InstructionMark im(this);
7662   prefixq(src, dst);
7663   emit_int8(0x3B);
7664   emit_operand(dst, src);
7665 }
7666 
7667 void Assembler::cmpxchgq(Register reg, Address adr) {
7668   InstructionMark im(this);
7669   prefixq(adr, reg);
7670   emit_int8(0x0F);
7671   emit_int8((unsigned char)0xB1);
7672   emit_operand(reg, adr);
7673 }
7674 
7675 void Assembler::cvtsi2sdq(XMMRegister dst, Register src) {
7676   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
7677   InstructionAttr attributes(AVX_128bit, /* rex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
7678   int encode = simd_prefix_and_encode(dst, dst, as_XMMRegister(src->encoding()), VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
7679   emit_int8(0x2A);
7680   emit_int8((unsigned char)(0xC0 | encode));
7681 }
7682 
7683 void Assembler::cvtsi2sdq(XMMRegister dst, Address src) {
7684   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
7685   InstructionMark im(this);
7686   InstructionAttr attributes(AVX_128bit, /* rex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
7687   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
7688   simd_prefix(dst, dst, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
7689   emit_int8(0x2A);
7690   emit_operand(dst, src);
7691 }
7692 
7693 void Assembler::cvtsi2ssq(XMMRegister dst, Address src) {
7694   NOT_LP64(assert(VM_Version::supports_sse(), ""));
7695   InstructionMark im(this);
7696   InstructionAttr attributes(AVX_128bit, /* rex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
7697   attributes.set_address_attributes(/* tuple_type */ EVEX_T1S, /* input_size_in_bits */ EVEX_64bit);
7698   simd_prefix(dst, dst, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
7699   emit_int8(0x2A);
7700   emit_operand(dst, src);
7701 }
7702 
7703 void Assembler::cvttsd2siq(Register dst, XMMRegister src) {
7704   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
7705   InstructionAttr attributes(AVX_128bit, /* rex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
7706   int encode = simd_prefix_and_encode(as_XMMRegister(dst->encoding()), xnoreg, src, VEX_SIMD_F2, VEX_OPCODE_0F, &attributes);
7707   emit_int8(0x2C);
7708   emit_int8((unsigned char)(0xC0 | encode));
7709 }
7710 
7711 void Assembler::cvttss2siq(Register dst, XMMRegister src) {
7712   NOT_LP64(assert(VM_Version::supports_sse(), ""));
7713   InstructionAttr attributes(AVX_128bit, /* rex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
7714   int encode = simd_prefix_and_encode(as_XMMRegister(dst->encoding()), xnoreg, src, VEX_SIMD_F3, VEX_OPCODE_0F, &attributes);
7715   emit_int8(0x2C);
7716   emit_int8((unsigned char)(0xC0 | encode));
7717 }
7718 
7719 void Assembler::decl(Register dst) {
7720   // Don't use it directly. Use MacroAssembler::decrementl() instead.
7721   // Use two-byte form (one-byte form is a REX prefix in 64-bit mode)
7722   int encode = prefix_and_encode(dst->encoding());
7723   emit_int8((unsigned char)0xFF);
7724   emit_int8((unsigned char)(0xC8 | encode));
7725 }
7726 
7727 void Assembler::decq(Register dst) {
7728   // Don't use it directly. Use MacroAssembler::decrementq() instead.
7729   // Use two-byte form (one-byte from is a REX prefix in 64-bit mode)
7730   int encode = prefixq_and_encode(dst->encoding());
7731   emit_int8((unsigned char)0xFF);
7732   emit_int8(0xC8 | encode);
7733 }
7734 
7735 void Assembler::decq(Address dst) {
7736   // Don't use it directly. Use MacroAssembler::decrementq() instead.
7737   InstructionMark im(this);
7738   prefixq(dst);
7739   emit_int8((unsigned char)0xFF);
7740   emit_operand(rcx, dst);
7741 }
7742 
7743 void Assembler::fxrstor(Address src) {
7744   prefixq(src);
7745   emit_int8(0x0F);
7746   emit_int8((unsigned char)0xAE);
7747   emit_operand(as_Register(1), src);
7748 }
7749 
7750 void Assembler::xrstor(Address src) {
7751   prefixq(src);
7752   emit_int8(0x0F);
7753   emit_int8((unsigned char)0xAE);
7754   emit_operand(as_Register(5), src);
7755 }
7756 
7757 void Assembler::fxsave(Address dst) {
7758   prefixq(dst);
7759   emit_int8(0x0F);
7760   emit_int8((unsigned char)0xAE);
7761   emit_operand(as_Register(0), dst);
7762 }
7763 
7764 void Assembler::xsave(Address dst) {
7765   prefixq(dst);
7766   emit_int8(0x0F);
7767   emit_int8((unsigned char)0xAE);
7768   emit_operand(as_Register(4), dst);
7769 }
7770 
7771 void Assembler::idivq(Register src) {
7772   int encode = prefixq_and_encode(src->encoding());
7773   emit_int8((unsigned char)0xF7);
7774   emit_int8((unsigned char)(0xF8 | encode));
7775 }
7776 
7777 void Assembler::imulq(Register dst, Register src) {
7778   int encode = prefixq_and_encode(dst->encoding(), src->encoding());
7779   emit_int8(0x0F);
7780   emit_int8((unsigned char)0xAF);
7781   emit_int8((unsigned char)(0xC0 | encode));
7782 }
7783 
7784 void Assembler::imulq(Register dst, Register src, int value) {
7785   int encode = prefixq_and_encode(dst->encoding(), src->encoding());
7786   if (is8bit(value)) {
7787     emit_int8(0x6B);
7788     emit_int8((unsigned char)(0xC0 | encode));
7789     emit_int8(value & 0xFF);
7790   } else {
7791     emit_int8(0x69);
7792     emit_int8((unsigned char)(0xC0 | encode));
7793     emit_int32(value);
7794   }
7795 }
7796 
7797 void Assembler::imulq(Register dst, Address src) {
7798   InstructionMark im(this);
7799   prefixq(src, dst);
7800   emit_int8(0x0F);
7801   emit_int8((unsigned char) 0xAF);
7802   emit_operand(dst, src);
7803 }
7804 
7805 void Assembler::incl(Register dst) {
7806   // Don't use it directly. Use MacroAssembler::incrementl() instead.
7807   // Use two-byte form (one-byte from is a REX prefix in 64-bit mode)
7808   int encode = prefix_and_encode(dst->encoding());
7809   emit_int8((unsigned char)0xFF);
7810   emit_int8((unsigned char)(0xC0 | encode));
7811 }
7812 
7813 void Assembler::incq(Register dst) {
7814   // Don't use it directly. Use MacroAssembler::incrementq() instead.
7815   // Use two-byte form (one-byte from is a REX prefix in 64-bit mode)
7816   int encode = prefixq_and_encode(dst->encoding());
7817   emit_int8((unsigned char)0xFF);
7818   emit_int8((unsigned char)(0xC0 | encode));
7819 }
7820 
7821 void Assembler::incq(Address dst) {
7822   // Don't use it directly. Use MacroAssembler::incrementq() instead.
7823   InstructionMark im(this);
7824   prefixq(dst);
7825   emit_int8((unsigned char)0xFF);
7826   emit_operand(rax, dst);
7827 }
7828 
7829 void Assembler::lea(Register dst, Address src) {
7830   leaq(dst, src);
7831 }
7832 
7833 void Assembler::leaq(Register dst, Address src) {
7834   InstructionMark im(this);
7835   prefixq(src, dst);
7836   emit_int8((unsigned char)0x8D);
7837   emit_operand(dst, src);
7838 }
7839 
7840 void Assembler::mov64(Register dst, int64_t imm64) {
7841   InstructionMark im(this);
7842   int encode = prefixq_and_encode(dst->encoding());
7843   emit_int8((unsigned char)(0xB8 | encode));
7844   emit_int64(imm64);
7845 }
7846 
7847 void Assembler::mov_literal64(Register dst, intptr_t imm64, RelocationHolder const& rspec) {
7848   InstructionMark im(this);
7849   int encode = prefixq_and_encode(dst->encoding());
7850   emit_int8(0xB8 | encode);
7851   emit_data64(imm64, rspec);
7852 }
7853 
7854 void Assembler::mov_narrow_oop(Register dst, int32_t imm32, RelocationHolder const& rspec) {
7855   InstructionMark im(this);
7856   int encode = prefix_and_encode(dst->encoding());
7857   emit_int8((unsigned char)(0xB8 | encode));
7858   emit_data((int)imm32, rspec, narrow_oop_operand);
7859 }
7860 
7861 void Assembler::mov_narrow_oop(Address dst, int32_t imm32,  RelocationHolder const& rspec) {
7862   InstructionMark im(this);
7863   prefix(dst);
7864   emit_int8((unsigned char)0xC7);
7865   emit_operand(rax, dst, 4);
7866   emit_data((int)imm32, rspec, narrow_oop_operand);
7867 }
7868 
7869 void Assembler::cmp_narrow_oop(Register src1, int32_t imm32, RelocationHolder const& rspec) {
7870   InstructionMark im(this);
7871   int encode = prefix_and_encode(src1->encoding());
7872   emit_int8((unsigned char)0x81);
7873   emit_int8((unsigned char)(0xF8 | encode));
7874   emit_data((int)imm32, rspec, narrow_oop_operand);
7875 }
7876 
7877 void Assembler::cmp_narrow_oop(Address src1, int32_t imm32, RelocationHolder const& rspec) {
7878   InstructionMark im(this);
7879   prefix(src1);
7880   emit_int8((unsigned char)0x81);
7881   emit_operand(rax, src1, 4);
7882   emit_data((int)imm32, rspec, narrow_oop_operand);
7883 }
7884 
7885 void Assembler::lzcntq(Register dst, Register src) {
7886   assert(VM_Version::supports_lzcnt(), "encoding is treated as BSR");
7887   emit_int8((unsigned char)0xF3);
7888   int encode = prefixq_and_encode(dst->encoding(), src->encoding());
7889   emit_int8(0x0F);
7890   emit_int8((unsigned char)0xBD);
7891   emit_int8((unsigned char)(0xC0 | encode));
7892 }
7893 
7894 void Assembler::movdq(XMMRegister dst, Register src) {
7895   // table D-1 says MMX/SSE2
7896   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
7897   InstructionAttr attributes(AVX_128bit, /* rex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
7898   int encode = simd_prefix_and_encode(dst, xnoreg, as_XMMRegister(src->encoding()), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
7899   emit_int8(0x6E);
7900   emit_int8((unsigned char)(0xC0 | encode));
7901 }
7902 
7903 void Assembler::movdq(Register dst, XMMRegister src) {
7904   // table D-1 says MMX/SSE2
7905   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
7906   InstructionAttr attributes(AVX_128bit, /* rex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
7907   // swap src/dst to get correct prefix
7908   int encode = simd_prefix_and_encode(src, xnoreg, as_XMMRegister(dst->encoding()), VEX_SIMD_66, VEX_OPCODE_0F, &attributes);
7909   emit_int8(0x7E);
7910   emit_int8((unsigned char)(0xC0 | encode));
7911 }
7912 
7913 void Assembler::movq(Register dst, Register src) {
7914   int encode = prefixq_and_encode(dst->encoding(), src->encoding());
7915   emit_int8((unsigned char)0x8B);
7916   emit_int8((unsigned char)(0xC0 | encode));
7917 }
7918 
7919 void Assembler::movq(Register dst, Address src) {
7920   InstructionMark im(this);
7921   prefixq(src, dst);
7922   emit_int8((unsigned char)0x8B);
7923   emit_operand(dst, src);
7924 }
7925 
7926 void Assembler::movq(Address dst, Register src) {
7927   InstructionMark im(this);
7928   prefixq(dst, src);
7929   emit_int8((unsigned char)0x89);
7930   emit_operand(src, dst);
7931 }
7932 
7933 void Assembler::movsbq(Register dst, Address src) {
7934   InstructionMark im(this);
7935   prefixq(src, dst);
7936   emit_int8(0x0F);
7937   emit_int8((unsigned char)0xBE);
7938   emit_operand(dst, src);
7939 }
7940 
7941 void Assembler::movsbq(Register dst, Register src) {
7942   int encode = prefixq_and_encode(dst->encoding(), src->encoding());
7943   emit_int8(0x0F);
7944   emit_int8((unsigned char)0xBE);
7945   emit_int8((unsigned char)(0xC0 | encode));
7946 }
7947 
7948 void Assembler::movslq(Register dst, int32_t imm32) {
7949   // dbx shows movslq(rcx, 3) as movq     $0x0000000049000000,(%rbx)
7950   // and movslq(r8, 3); as movl     $0x0000000048000000,(%rbx)
7951   // as a result we shouldn't use until tested at runtime...
7952   ShouldNotReachHere();
7953   InstructionMark im(this);
7954   int encode = prefixq_and_encode(dst->encoding());
7955   emit_int8((unsigned char)(0xC7 | encode));
7956   emit_int32(imm32);
7957 }
7958 
7959 void Assembler::movslq(Address dst, int32_t imm32) {
7960   assert(is_simm32(imm32), "lost bits");
7961   InstructionMark im(this);
7962   prefixq(dst);
7963   emit_int8((unsigned char)0xC7);
7964   emit_operand(rax, dst, 4);
7965   emit_int32(imm32);
7966 }
7967 
7968 void Assembler::movslq(Register dst, Address src) {
7969   InstructionMark im(this);
7970   prefixq(src, dst);
7971   emit_int8(0x63);
7972   emit_operand(dst, src);
7973 }
7974 
7975 void Assembler::movslq(Register dst, Register src) {
7976   int encode = prefixq_and_encode(dst->encoding(), src->encoding());
7977   emit_int8(0x63);
7978   emit_int8((unsigned char)(0xC0 | encode));
7979 }
7980 
7981 void Assembler::movswq(Register dst, Address src) {
7982   InstructionMark im(this);
7983   prefixq(src, dst);
7984   emit_int8(0x0F);
7985   emit_int8((unsigned char)0xBF);
7986   emit_operand(dst, src);
7987 }
7988 
7989 void Assembler::movswq(Register dst, Register src) {
7990   int encode = prefixq_and_encode(dst->encoding(), src->encoding());
7991   emit_int8((unsigned char)0x0F);
7992   emit_int8((unsigned char)0xBF);
7993   emit_int8((unsigned char)(0xC0 | encode));
7994 }
7995 
7996 void Assembler::movzbq(Register dst, Address src) {
7997   InstructionMark im(this);
7998   prefixq(src, dst);
7999   emit_int8((unsigned char)0x0F);
8000   emit_int8((unsigned char)0xB6);
8001   emit_operand(dst, src);
8002 }
8003 
8004 void Assembler::movzbq(Register dst, Register src) {
8005   int encode = prefixq_and_encode(dst->encoding(), src->encoding());
8006   emit_int8(0x0F);
8007   emit_int8((unsigned char)0xB6);
8008   emit_int8(0xC0 | encode);
8009 }
8010 
8011 void Assembler::movzwq(Register dst, Address src) {
8012   InstructionMark im(this);
8013   prefixq(src, dst);
8014   emit_int8((unsigned char)0x0F);
8015   emit_int8((unsigned char)0xB7);
8016   emit_operand(dst, src);
8017 }
8018 
8019 void Assembler::movzwq(Register dst, Register src) {
8020   int encode = prefixq_and_encode(dst->encoding(), src->encoding());
8021   emit_int8((unsigned char)0x0F);
8022   emit_int8((unsigned char)0xB7);
8023   emit_int8((unsigned char)(0xC0 | encode));
8024 }
8025 
8026 void Assembler::mulq(Address src) {
8027   InstructionMark im(this);
8028   prefixq(src);
8029   emit_int8((unsigned char)0xF7);
8030   emit_operand(rsp, src);
8031 }
8032 
8033 void Assembler::mulq(Register src) {
8034   int encode = prefixq_and_encode(src->encoding());
8035   emit_int8((unsigned char)0xF7);
8036   emit_int8((unsigned char)(0xE0 | encode));
8037 }
8038 
8039 void Assembler::mulxq(Register dst1, Register dst2, Register src) {
8040   assert(VM_Version::supports_bmi2(), "bit manipulation instructions not supported");
8041   InstructionAttr attributes(AVX_128bit, /* vex_w */ true, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
8042   int encode = vex_prefix_and_encode(dst1->encoding(), dst2->encoding(), src->encoding(), VEX_SIMD_F2, VEX_OPCODE_0F_38, &attributes);
8043   emit_int8((unsigned char)0xF6);
8044   emit_int8((unsigned char)(0xC0 | encode));
8045 }
8046 
8047 void Assembler::negq(Register dst) {
8048   int encode = prefixq_and_encode(dst->encoding());
8049   emit_int8((unsigned char)0xF7);
8050   emit_int8((unsigned char)(0xD8 | encode));
8051 }
8052 
8053 void Assembler::notq(Register dst) {
8054   int encode = prefixq_and_encode(dst->encoding());
8055   emit_int8((unsigned char)0xF7);
8056   emit_int8((unsigned char)(0xD0 | encode));
8057 }
8058 
8059 void Assembler::orq(Address dst, int32_t imm32) {
8060   InstructionMark im(this);
8061   prefixq(dst);
8062   emit_int8((unsigned char)0x81);
8063   emit_operand(rcx, dst, 4);
8064   emit_int32(imm32);
8065 }
8066 
8067 void Assembler::orq(Register dst, int32_t imm32) {
8068   (void) prefixq_and_encode(dst->encoding());
8069   emit_arith(0x81, 0xC8, dst, imm32);
8070 }
8071 
8072 void Assembler::orq(Register dst, Address src) {
8073   InstructionMark im(this);
8074   prefixq(src, dst);
8075   emit_int8(0x0B);
8076   emit_operand(dst, src);
8077 }
8078 
8079 void Assembler::orq(Register dst, Register src) {
8080   (void) prefixq_and_encode(dst->encoding(), src->encoding());
8081   emit_arith(0x0B, 0xC0, dst, src);
8082 }
8083 
8084 void Assembler::popa() { // 64bit
8085   movq(r15, Address(rsp, 0));
8086   movq(r14, Address(rsp, wordSize));
8087   movq(r13, Address(rsp, 2 * wordSize));
8088   movq(r12, Address(rsp, 3 * wordSize));
8089   movq(r11, Address(rsp, 4 * wordSize));
8090   movq(r10, Address(rsp, 5 * wordSize));
8091   movq(r9,  Address(rsp, 6 * wordSize));
8092   movq(r8,  Address(rsp, 7 * wordSize));
8093   movq(rdi, Address(rsp, 8 * wordSize));
8094   movq(rsi, Address(rsp, 9 * wordSize));
8095   movq(rbp, Address(rsp, 10 * wordSize));
8096   // skip rsp
8097   movq(rbx, Address(rsp, 12 * wordSize));
8098   movq(rdx, Address(rsp, 13 * wordSize));
8099   movq(rcx, Address(rsp, 14 * wordSize));
8100   movq(rax, Address(rsp, 15 * wordSize));
8101 
8102   addq(rsp, 16 * wordSize);
8103 }
8104 
8105 void Assembler::popcntq(Register dst, Address src) {
8106   assert(VM_Version::supports_popcnt(), "must support");
8107   InstructionMark im(this);
8108   emit_int8((unsigned char)0xF3);
8109   prefixq(src, dst);
8110   emit_int8((unsigned char)0x0F);
8111   emit_int8((unsigned char)0xB8);
8112   emit_operand(dst, src);
8113 }
8114 
8115 void Assembler::popcntq(Register dst, Register src) {
8116   assert(VM_Version::supports_popcnt(), "must support");
8117   emit_int8((unsigned char)0xF3);
8118   int encode = prefixq_and_encode(dst->encoding(), src->encoding());
8119   emit_int8((unsigned char)0x0F);
8120   emit_int8((unsigned char)0xB8);
8121   emit_int8((unsigned char)(0xC0 | encode));
8122 }
8123 
8124 void Assembler::popq(Address dst) {
8125   InstructionMark im(this);
8126   prefixq(dst);
8127   emit_int8((unsigned char)0x8F);
8128   emit_operand(rax, dst);
8129 }
8130 
8131 void Assembler::pusha() { // 64bit
8132   // we have to store original rsp.  ABI says that 128 bytes
8133   // below rsp are local scratch.
8134   movq(Address(rsp, -5 * wordSize), rsp);
8135 
8136   subq(rsp, 16 * wordSize);
8137 
8138   movq(Address(rsp, 15 * wordSize), rax);
8139   movq(Address(rsp, 14 * wordSize), rcx);
8140   movq(Address(rsp, 13 * wordSize), rdx);
8141   movq(Address(rsp, 12 * wordSize), rbx);
8142   // skip rsp
8143   movq(Address(rsp, 10 * wordSize), rbp);
8144   movq(Address(rsp, 9 * wordSize), rsi);
8145   movq(Address(rsp, 8 * wordSize), rdi);
8146   movq(Address(rsp, 7 * wordSize), r8);
8147   movq(Address(rsp, 6 * wordSize), r9);
8148   movq(Address(rsp, 5 * wordSize), r10);
8149   movq(Address(rsp, 4 * wordSize), r11);
8150   movq(Address(rsp, 3 * wordSize), r12);
8151   movq(Address(rsp, 2 * wordSize), r13);
8152   movq(Address(rsp, wordSize), r14);
8153   movq(Address(rsp, 0), r15);
8154 }
8155 
8156 void Assembler::pushq(Address src) {
8157   InstructionMark im(this);
8158   prefixq(src);
8159   emit_int8((unsigned char)0xFF);
8160   emit_operand(rsi, src);
8161 }
8162 
8163 void Assembler::rclq(Register dst, int imm8) {
8164   assert(isShiftCount(imm8 >> 1), "illegal shift count");
8165   int encode = prefixq_and_encode(dst->encoding());
8166   if (imm8 == 1) {
8167     emit_int8((unsigned char)0xD1);
8168     emit_int8((unsigned char)(0xD0 | encode));
8169   } else {
8170     emit_int8((unsigned char)0xC1);
8171     emit_int8((unsigned char)(0xD0 | encode));
8172     emit_int8(imm8);
8173   }
8174 }
8175 
8176 void Assembler::rcrq(Register dst, int imm8) {
8177   assert(isShiftCount(imm8 >> 1), "illegal shift count");
8178   int encode = prefixq_and_encode(dst->encoding());
8179   if (imm8 == 1) {
8180     emit_int8((unsigned char)0xD1);
8181     emit_int8((unsigned char)(0xD8 | encode));
8182   } else {
8183     emit_int8((unsigned char)0xC1);
8184     emit_int8((unsigned char)(0xD8 | encode));
8185     emit_int8(imm8);
8186   }
8187 }
8188 
8189 void Assembler::rorq(Register dst, int imm8) {
8190   assert(isShiftCount(imm8 >> 1), "illegal shift count");
8191   int encode = prefixq_and_encode(dst->encoding());
8192   if (imm8 == 1) {
8193     emit_int8((unsigned char)0xD1);
8194     emit_int8((unsigned char)(0xC8 | encode));
8195   } else {
8196     emit_int8((unsigned char)0xC1);
8197     emit_int8((unsigned char)(0xc8 | encode));
8198     emit_int8(imm8);
8199   }
8200 }
8201 
8202 void Assembler::rorxq(Register dst, Register src, int imm8) {
8203   assert(VM_Version::supports_bmi2(), "bit manipulation instructions not supported");
8204   InstructionAttr attributes(AVX_128bit, /* vex_w */ true, /* legacy_mode */ true, /* no_mask_reg */ false, /* uses_vl */ false);
8205   int encode = vex_prefix_and_encode(dst->encoding(), 0, src->encoding(), VEX_SIMD_F2, VEX_OPCODE_0F_3A, &attributes);
8206   emit_int8((unsigned char)0xF0);
8207   emit_int8((unsigned char)(0xC0 | encode));
8208   emit_int8(imm8);
8209 }
8210 
8211 void Assembler::sarq(Register dst, int imm8) {
8212   assert(isShiftCount(imm8 >> 1), "illegal shift count");
8213   int encode = prefixq_and_encode(dst->encoding());
8214   if (imm8 == 1) {
8215     emit_int8((unsigned char)0xD1);
8216     emit_int8((unsigned char)(0xF8 | encode));
8217   } else {
8218     emit_int8((unsigned char)0xC1);
8219     emit_int8((unsigned char)(0xF8 | encode));
8220     emit_int8(imm8);
8221   }
8222 }
8223 
8224 void Assembler::sarq(Register dst) {
8225   int encode = prefixq_and_encode(dst->encoding());
8226   emit_int8((unsigned char)0xD3);
8227   emit_int8((unsigned char)(0xF8 | encode));
8228 }
8229 
8230 void Assembler::sbbq(Address dst, int32_t imm32) {
8231   InstructionMark im(this);
8232   prefixq(dst);
8233   emit_arith_operand(0x81, rbx, dst, imm32);
8234 }
8235 
8236 void Assembler::sbbq(Register dst, int32_t imm32) {
8237   (void) prefixq_and_encode(dst->encoding());
8238   emit_arith(0x81, 0xD8, dst, imm32);
8239 }
8240 
8241 void Assembler::sbbq(Register dst, Address src) {
8242   InstructionMark im(this);
8243   prefixq(src, dst);
8244   emit_int8(0x1B);
8245   emit_operand(dst, src);
8246 }
8247 
8248 void Assembler::sbbq(Register dst, Register src) {
8249   (void) prefixq_and_encode(dst->encoding(), src->encoding());
8250   emit_arith(0x1B, 0xC0, dst, src);
8251 }
8252 
8253 void Assembler::shlq(Register dst, int imm8) {
8254   assert(isShiftCount(imm8 >> 1), "illegal shift count");
8255   int encode = prefixq_and_encode(dst->encoding());
8256   if (imm8 == 1) {
8257     emit_int8((unsigned char)0xD1);
8258     emit_int8((unsigned char)(0xE0 | encode));
8259   } else {
8260     emit_int8((unsigned char)0xC1);
8261     emit_int8((unsigned char)(0xE0 | encode));
8262     emit_int8(imm8);
8263   }
8264 }
8265 
8266 void Assembler::shlq(Register dst) {
8267   int encode = prefixq_and_encode(dst->encoding());
8268   emit_int8((unsigned char)0xD3);
8269   emit_int8((unsigned char)(0xE0 | encode));
8270 }
8271 
8272 void Assembler::shrq(Register dst, int imm8) {
8273   assert(isShiftCount(imm8 >> 1), "illegal shift count");
8274   int encode = prefixq_and_encode(dst->encoding());
8275   emit_int8((unsigned char)0xC1);
8276   emit_int8((unsigned char)(0xE8 | encode));
8277   emit_int8(imm8);
8278 }
8279 
8280 void Assembler::shrq(Register dst) {
8281   int encode = prefixq_and_encode(dst->encoding());
8282   emit_int8((unsigned char)0xD3);
8283   emit_int8(0xE8 | encode);
8284 }
8285 
8286 void Assembler::subq(Address dst, int32_t imm32) {
8287   InstructionMark im(this);
8288   prefixq(dst);
8289   emit_arith_operand(0x81, rbp, dst, imm32);
8290 }
8291 
8292 void Assembler::subq(Address dst, Register src) {
8293   InstructionMark im(this);
8294   prefixq(dst, src);
8295   emit_int8(0x29);
8296   emit_operand(src, dst);
8297 }
8298 
8299 void Assembler::subq(Register dst, int32_t imm32) {
8300   (void) prefixq_and_encode(dst->encoding());
8301   emit_arith(0x81, 0xE8, dst, imm32);
8302 }
8303 
8304 // Force generation of a 4 byte immediate value even if it fits into 8bit
8305 void Assembler::subq_imm32(Register dst, int32_t imm32) {
8306   (void) prefixq_and_encode(dst->encoding());
8307   emit_arith_imm32(0x81, 0xE8, dst, imm32);
8308 }
8309 
8310 void Assembler::subq(Register dst, Address src) {
8311   InstructionMark im(this);
8312   prefixq(src, dst);
8313   emit_int8(0x2B);
8314   emit_operand(dst, src);
8315 }
8316 
8317 void Assembler::subq(Register dst, Register src) {
8318   (void) prefixq_and_encode(dst->encoding(), src->encoding());
8319   emit_arith(0x2B, 0xC0, dst, src);
8320 }
8321 
8322 void Assembler::testq(Register dst, int32_t imm32) {
8323   // not using emit_arith because test
8324   // doesn't support sign-extension of
8325   // 8bit operands
8326   int encode = dst->encoding();
8327   if (encode == 0) {
8328     prefix(REX_W);
8329     emit_int8((unsigned char)0xA9);
8330   } else {
8331     encode = prefixq_and_encode(encode);
8332     emit_int8((unsigned char)0xF7);
8333     emit_int8((unsigned char)(0xC0 | encode));
8334   }
8335   emit_int32(imm32);
8336 }
8337 
8338 void Assembler::testq(Register dst, Register src) {
8339   (void) prefixq_and_encode(dst->encoding(), src->encoding());
8340   emit_arith(0x85, 0xC0, dst, src);
8341 }
8342 
8343 void Assembler::xaddq(Address dst, Register src) {
8344   InstructionMark im(this);
8345   prefixq(dst, src);
8346   emit_int8(0x0F);
8347   emit_int8((unsigned char)0xC1);
8348   emit_operand(src, dst);
8349 }
8350 
8351 void Assembler::xchgq(Register dst, Address src) {
8352   InstructionMark im(this);
8353   prefixq(src, dst);
8354   emit_int8((unsigned char)0x87);
8355   emit_operand(dst, src);
8356 }
8357 
8358 void Assembler::xchgq(Register dst, Register src) {
8359   int encode = prefixq_and_encode(dst->encoding(), src->encoding());
8360   emit_int8((unsigned char)0x87);
8361   emit_int8((unsigned char)(0xc0 | encode));
8362 }
8363 
8364 void Assembler::xorq(Register dst, Register src) {
8365   (void) prefixq_and_encode(dst->encoding(), src->encoding());
8366   emit_arith(0x33, 0xC0, dst, src);
8367 }
8368 
8369 void Assembler::xorq(Register dst, Address src) {
8370   InstructionMark im(this);
8371   prefixq(src, dst);
8372   emit_int8(0x33);
8373   emit_operand(dst, src);
8374 }
8375 
8376 #endif // !LP64